63
C OPYRIGHT 2015
BY
T HE J OURNAL
OF
B ONE
AND J OINT
S URGERY, I NCORPORATED
Synovial Fluid Interleukin-6 as a Predictor
of Periprosthetic Shoulder Infection
Salvatore J. Frangiamore, MD, Anas Saleh, MD, Mario Farias Kovac, MD, Matthew J. Grosso, BS, Xiaochun Zhang, MD,
Thomas W. Bauer, MD, Thomas M. Daly, MD, Eric T. Ricchetti, MD, and Joseph P. Iannotti, MD, PhD
Investigation performed at the Orthopaedic and Rheumatologic Institute, Cleveland Clinic Foundation, Cleveland, Ohio
Background: Diagnosis of periprosthetic joint infection (PJI) in patients undergoing revision shoulder arthroplasty is
challenging because of the low virulence of the most common infecting organisms. The goal of this study was to evaluate
the diagnostic utility of measuring synovial fluid interleukin-6 (IL-6) levels for identifying PJI of the shoulder.
Methods: Thirty-two consecutive patients evaluated for pain at the site of a shoulder arthroplasty were prospectively enrolled
from November 2012 to September 2013 and underwent revision surgery (thirty-five procedures during which samples were
obtained for synovial fluid IL-6 analysis). Cases were categorized into infection (n = 15) and no-infection (n = 20) groups on the
basis of objective preoperative and intraoperative findings. Twenty patients treated with arthroscopic rotator cuff repair were
also enrolled to serve as a non-infected control group. Synovial fluid was obtained through aspiration intraoperatively for all
patients, as well as preoperatively for some. Synovial fluid IL-6 levels were measured with use of a cytokine immunoassay that
utilizes electrochemiluminescent detection. A receiver operating characteristic curve was used to determine the diagnostic
utility of synovial fluid IL-6 analysis.
Results: Based on receiver operating characteristic curve analysis, synovial fluid IL-6 measurement had an area under
the curve of 0.891 with an ideal cutoff value of 359.3 pg/mL. The sensitivity, specificity, and positive and negative
likelihood ratios were 87%, 90%, 8.45, and 0.15, respectively. Seven patients who underwent a single-stage revision had
negative results on standard perioperative testing, including the erythrocyte sedimentation rate and C-reactive protein
levels, but multiple positive intraoperative tissue cultures. The level of synovial fluid IL-6 was elevated in five of these
seven patients, with a median value of 1400 pg/mL. Intraoperative synovial fluid IL-6 values correlated well with preoperative IL-6 synovial fluid values (correlation = 0.61; p = 0.025) and frozen-section histologic findings (p < 0.001).
Synovial fluid IL-6 levels were also significantly elevated in patients with Propionibacterium acnes infection (p = 0.01).
Conclusions: Measurement of synovial fluid IL-6 levels is more sensitive and specific than current preoperative testing
for predicting positive cultures for patients undergoing revision shoulder arthroplasty. This diagnostic accuracy can lead to
improved decision-making in the management of PJI.
Level of Evidence: Diagnostic Level II. See Instructions for Authors for a complete description of levels of evidence.
Peer Review: This article was reviewed by the Editor-in-Chief and one Deputy Editor, and it underwent blinded review by two or more outside experts. It was also reviewed
by an expert in methodology and statistics. The Deputy Editor reviewed each revision of the article, and it underwent a final review by the Editor-in-Chief prior to publication.
Final corrections and clarifications occurred during one or more exchanges between the author(s) and copyeditors.
O
ne of the most serious complications following shoulder arthroplasty is periprosthetic joint infection (PJI),
with a reported prevalence between 0.7% and 4%1-3.
PJI represents 2.9% to 4.6% of all complications of shoulder
arthroplasty, and the prevalence is higher following revision
surgery and reverse total shoulder arthroplasty1-5. Chronic infections may be difficult to eradicate and lead to poor outcomes1,6. Correct preoperative diagnosis is important when
Disclosure: None of the authors received payments or services, either directly or indirectly (i.e., via his or her institution), from a third party in support of
any aspect of this work. One or more of the authors, or his or her institution, has had a financial relationship, in the thirty-six months prior to submission of
this work, with an entity in the biomedical arena that could be perceived to influence or have the potential to influence what is written in this work. Also, one
or more of the authors has had another relationship, or has engaged in another activity, that could be perceived to influence or have the potential to
influence what is written in this work. The complete Disclosures of Potential Conflicts of Interest submitted by authors are always provided with the
online version of the article.
J Bone Joint Surg Am. 2015;97:63-70
d
http://dx.doi.org/10.2106/JBJS.N.00104
64
TH E JO U R NA L O F B O N E & JO I N T SU RG E RY J B J S . O RG
V O L U M E 97-A N U M B E R 1 J A N UA RY 7, 2 015
d
d
d
deciding on a single-stage or two-stage exchange arthroplasty
in patients with chronic PJI. However, PJI in the shoulder often
presents a diagnostic challenge due to the indolent nature of
the common infecting organisms, including Propionibacterium
acnes (P. acnes)7 and the lower efficacy of common diagnostic
markers relative to those for periprosthetic hip and knee infections. The sensitivities of the serum erythrocyte sedimentation
rate (ESR) and C-reactive protein (CRP) level in the diagnosis of
PJI in the shoulder have been reported to be 16% and 42%,
respectively8, compared with 95% and 88% for the diagnosis of
hip PJI and 94% and 97% for the diagnosis of knee PJI9. As a
result, it is common for a one-stage exchange to be performed
because perioperative markers of infection were negative at the
time of the revision surgery, only to be followed by an unexpected
positive result of intraoperative cultures after the surgery.
Newer tests for the diagnosis of PJI have recently been
studied and may be more efficacious for the shoulder. In
particular, the measurement of pro-inflammatory cytokine
interleukin-6 (IL-6) levels may hold promise, with multiple
studies showing both synovial fluid10 and serum 11,12 levels to
have sensitivities and specificities of >95% in the diagnosis
of hip and knee PJI. Although one small study showed a benefit
of measuring serum IL-6 levels in the shoulder13, subsequent
studies examining primarily indolent infections have shown
analysis of serum IL-6 levels to have poor sensitivity and minimum diagnostic value in the diagnosis of PJI of the shoulder14,15.
We are not aware of any studies evaluating the efficacy of synovial
Fig. 1
Flow diagram showing excluded and included patients.
S Y N O V I A L F L U I D I N T E R L E U K I N -6 A S A P R E D I C T O R
PER IPROSTHETIC SHOULDER INFECTION
OF
fluid IL-6 analysis for diagnosis of PJI of the shoulder. Therefore, the goal of this study was to evaluate the diagnostic utility
of synovial fluid IL-6 analysis for patients undergoing revision
shoulder arthroplasty.
Materials and Methods
Patient Selection
O
ur institutional review board approved this study, and informed consent
was obtained from all patients. All patients, treated by two shoulder
surgeons (J.P.I. and E.T.R.), who were evaluated for pain at the site of a shoulder
arthroplasty were prospectively enrolled from November 2012 to September
2013. Forty-six consecutive patients were enrolled in the study. All underwent
a preoperative routine workup for PJI, including measurements of the serum
ESR and CRP level as well as shoulder aspiration for synovial fluid culture and
IL-6 analysis. Patients who did not subsequently undergo revision shoulder
arthroplasty because there was not a clear indication for surgical intervention,
because their pain resolved, or because they preferred a nonoperative approach
(n = 11) were excluded, as intraoperative tissue specimens for cultures and
frozen-section histologic analysis were required for the diagnosis of PJI. The
median synovial fluid IL-6 level in these excluded patients was 305 pg/mL
(interquartile range, 81 to 460 pg/mL), with four having an elevated level. Three
of the forty-six patients had a dry arthrocentesis and synovial fluid could not be
obtained either intraoperatively or preoperatively for IL-6 analysis; they were
therefore excluded as well. There were no differences in clinical or laboratory
parameters between these fourteen excluded patients and the included cohort.
Thirty-two patients underwent revision shoulder arthroplasty, during which
synovial fluid was obtained, and synovial fluid was also obtained preoperatively
for IL-6 analysis for nine of them (Fig. 1). Samples from thirty-five procedures
were analyzed as three patients had samples obtained both at the time of their
initial revision surgery and at the second-stage reimplantation. All IL-6 synovial
65
TH E JO U R NA L O F B O N E & JO I N T SU RG E RY J B J S . O RG
V O L U M E 97-A N U M B E R 1 J A N UA RY 7, 2 015
d
d
d
TABLE I Clinical Characteristics of Patients Undergoing
Revision Shoulder Arthroplasty
Variable
Age* (yr)
Male sex†
60.7 ± 13.2
15 (46.9)
ESR‡ (mm/hr)
8 (2.2, 14.2)
CPR level‡ (mg/dL)
0.3 (0.1, 0.7)
Initial prosthesis†
Anatomic total shoulder
14 (40.0)
Hemiarthroplasty
15 (42.9)
Reverse shoulder
2 (5.7)
Spacer
4 (11.4)
Revision prosthesis†
Anatomic total shoulder
6 (17.1)
Hemiarthroplasty
4 (11.4)
Reverse shoulder
16 (45.7)
Spacer
9 (25.7)
Isolated microorganism†
Coagulase-negative Staphylococcus
6 (17.1)
Methicillin-resistant Staphylococcus
aureus
P. acnes
1 (2.9)
Others
2 (5.7)
None
9 (25.7)
19 (54.3)
Number of positive cultures†
0
19 (54.3)
1
2 (5.7)
>1
14 (40.0)
Positive frozen section†
S Y N O V I A L F L U I D I N T E R L E U K I N -6 A S A P R E D I C T O R
PER IPROSTHETIC SHOULDER INFECTION
the decision to use a standard or reverse prosthesis based on the amount of
bone loss and rotator cuff deficiency. Two-stage exchange included removal of
all components and placement of an antibiotic spacer during the first stage, and
reimplantation of a standard or reverse prosthesis at the second surgery based
on the amount of bone loss and rotator cuff deficiency.
To obtain baseline data for non-arthritic shoulders, we measured synovial fluid IL-6 levels in twenty patients undergoing primary arthroscopic rotator cuff repair who had no evidence of arthritic changes. Samples from
primary arthroplasties were not used for controls because of inconsistency in
16,17
the literature describing IL-6 levels in patients with osteoarthritis
. Rotator
cuff tears were chosen as a condition not caused by infection and in which
adequate intraoperative synovial fluid samples could be obtained for analysis.
Thirty consecutive patients undergoing arthroscopic rotator cuff repair had
been enrolled in the study to obtain synovial fluid IL-6 levels, but no fluid could
be obtained from ten of them (Fig. 1).
Synovial Fluid Handling and IL-6 Testing
For IL-6 testing, synovial fluid samples were centrifuged at 2000 rpm for ten
minutes, within two hours after collection, to remove all cellular and particulate
content. The resulting supernatant was divided into aliquots and stored in a 270C
freezer until samples were sent for testing. IL-6 levels in synovial fluid were
measured with use of the Human Proinflammatory Ultra-Sensitive Kit on the
Meso Scale Discovery (MSD) MULTI-ARRAY platform (Rockville, Maryland)
(see Appendix).
Infection Criteria
There is no gold standard to establish or rule out the diagnosis of infection at the
site of a shoulder arthroplasty. There is controversy even when intraoperative
TABLE II Periprosthetic Shoulder Infection Criteria
Category
Definite infection
8 (25.8)
Criteria*
At least one positive preoperative or
intraoperative finding of infection
and multiple positive intraoperative
cultures with the same organism
OR
*Mean and standard deviation. †Number (percent). ‡Median (25th
percentile, 75th percentile).
fluid analysis was performed in a delayed manner with use of the research
protocol described below. Therefore, these data were not available for clinical
decision-making either preoperatively or intraoperatively and were analyzed
strictly for the purposes of this study.
At the time of the revision surgery, prior to arthrotomy, synovial fluid
was obtained for culture and IL-6 analysis via direct needle aspiration to mimic
preoperative conditions and to avoid contamination with blood. Multiple
periprosthetic tissue specimens were also obtained intraoperatively from all
patients (mean, four cultures; standard deviation [SD], 1.5; range, two to
seven). These included samples along the glenoid and/or humeral components,
intramedullary tissue if a humeral stem was removed, tissue off the glenoid
osseous surface and glenoid holes if a glenoid component was removed, and
deep capsular tissue or pseudocapsular tissue in a cuff-deficient revision. Fluid
and tissue specimens were processed according to standard laboratory protocol
with cultures maintained for fourteen days (see Appendix). Intraoperative
tissue specimens also underwent frozen-section histologic analysis, with five
or more polymorphonuclear leukocytes in each of three or more high-power
(·400) fields used as the institutional criterion for acute inflammation consistent with infection. Basic demographic and perioperative characteristics of
the revision surgery cohort are shown in Table I. Patients undergoing singlestage exchange had revision of the humeral and/or glenoid components, with
OF
One positive preoperative (aspirate)
culture and one positive intraoperative
culture with the same organism
Probable infection
At least one positive preoperative or
intraoperative finding of infection and
one positive intraoperative culture
OR
No preoperative or intraoperative
findings of infection and multiple
positive intraoperative cultures
with the same organism
Probable contaminant
No preoperative or intraoperative
findings of infection and one
positive intraoperative culture
No evidence for
infection
No preoperative or intraoperative
findings of infection and no positive
intraoperative cultures
*Preoperative or intraoperative findings of infection: preoperative
clinical signs (swelling, sinus tract, redness, drainage), positive
ESR or CRP value, intraoperative gross findings (purulent drainage,
necrosis), positive intraoperative frozen section.
66
TH E JO U R NA L O F B O N E & JO I N T SU RG E RY J B J S . O RG
V O L U M E 97-A N U M B E R 1 J A N UA RY 7, 2 015
d
d
d
S Y N O V I A L F L U I D I N T E R L E U K I N -6 A S A P R E D I C T O R
PER IPROSTHETIC SHOULDER INFECTION
OF
Results
f the thirty-five revision shoulder arthroplasties, fifteen
(43%) met the criteria for the infection group (ten probable
infection and five definite infection). The no-infection group,
therefore, included twenty revision shoulder arthroplasties (one
probable contamination and nineteen no evidence of infection)
and twenty arthroscopic rotator cuff repairs. Synovial fluid IL-6
analysis demonstrated high diagnostic accuracy for PJI with an
area under the curve of 0.891 and an ideal cutoff value (determined with use of Youden’s criterion) of 359.3 pg/mL (Fig. 2).
Synovial fluid IL-6 analysis had high sensitivity and specificity
(87% and 90%, respectively), with positive and negative likelihood ratios of 8.45 and 0.15, respectively. A positive likelihood
ratio of 8.45 indicates that a positive IL-6 value increases the
pretest odds of infection by a factor of 8.5. Similarly, a negative
likelihood ratio of 0.15 indicates that a negative IL-6 value decreases the pretest odds of infection by a factor of 0.15. When
applying the Musculoskeletal Infection Society (MSIS) definition
for PJI instead of our proposed criteria, the area under the curve
was 0.892 and the ideal cutoff for IL-6 was 446 pg/mL. The
sensitivity, specificity, and positive and negative likelihood ratios
were 86%, 95%, 17.2, and 0.15, respectively.
With use of the cutoff value of 359.3 pg/mL for the diagnosis, synovial fluid IL-6 analysis resulted in four false-positive
and two false-negative results (Fig. 3). Of the four false-positive
results, two were for patients who underwent arthroscopic
rotator cuff repair, one was for a sample obtained during a
O
Fig. 2
Receiver operating characteristic curve for prediction of periprosthetic
shoulder infection based on IL-6 levels. AUC = area under the curve.
cultures are positive because of the difficulty in distinguishing true infection
from contamination, especially with less virulent organisms such as P. acnes.
Therefore, we used a spectrum of infection categories based on objective clinical,
laboratory, and histologic criteria, including serum ESR and CRP level, frozensection histologic analysis, and results of cultures of preoperative synovial fluid
aspirate and intraoperative tissue and synovial fluid specimens (Table II). These
3,18-20
criteria were consistent with those described in the current literature
. To
facilitate the analysis of the diagnostic utility of synovial fluid IL-6 results, the
subcategories were condensed into two groups: a no-infection group and an
infection group. The no-infection group included patients treated with revision
who met the criteria for the probable-contamination category or no evidence of
infection (Table II) and control patients who underwent arthroscopic rotator
cuff repair. The infection group included patients who met the criteria for definite
infection or probable infection (Table II). Most of the literature defines patients
who meet the criteria for our probable-contamination group as not infected, which
3,21,22
is why these patients were included in the no-infection group
. The diagnosis
category for all patients was established by authors (J.P.I., E.T.R., T.W.B., S.J.F., and
M.J.G.) blinded to the IL-6 values.
Data Analysis
The ability of synovial fluid IL-6 analysis to predict infection status was explored by using receiver operating characteristic curves. Youden’s criterion was
used to choose an ideal cutoff point for determining infection status on the
basis of the IL-6 level. According to this criterion, the point on the receiver
operating characteristic curve that maximizes the sum of sensitivity and
specificity is considered to be the ideal cutoff level. The efficacy of using IL-6
analysis to predict infection status was described by using the area under the
curve, sensitivity, specificity, and positive and negative likelihood ratios. Relationships between intraoperative synovial fluid IL-6 values and other objective
variables were also analyzed, with significance set at p < 0.05 (see Appendix).
Source of Funding
This study was funded by the Orthopaedic Operating Room of the Future Grant
in the Orthopaedic and Rheumatologic Institute at the Cleveland Clinic.
Fig. 3
Plot of IL-6 levels by clinical diagnosis. The horizontal line in the large box is
the chosen cutoff value of 359.3 pg/mL for the IL-6 level that indicates
infection, the horizontal lines in the three smaller boxes are the median IL-6
level, the I bars are the standard deviation, and the top and bottom of the
boxes are the 25th and 75th percentiles. Note that the IL-6 levels are
plotted on a logarithmic scale.
67
TH E JO U R NA L O F B O N E & JO I N T SU RG E RY J B J S . O RG
V O L U M E 97-A N U M B E R 1 J A N UA RY 7, 2 015
d
d
d
TABLE III Correlations with Intraoperative IL-6 Levels
Factor
Correlation
P Value*
Preoperative IL-6 level
0.61
0.025
Serum white blood-cell count
0.09
0.46
CRP level
0.29
0.031
ESR
0.17
0.21
Number of positive intraoperative
cultures
0.58
<0.001
20.36
0.061
Days to culture positivity
*Kendall tau rank correlation.
second-stage reimplantation that was done within four weeks
after the first-stage procedure, and one was for a case classified
as probable contamination with only one positive intraoperative
culture (for Staphylococcus epidermidis). Both patients with a
false-negative result had had multiple operations prior to their
revision and had cultures that were positive for P. acnes. They
were treated with one-stage revision and started on intravenous
(IV) antibiotics when the cultures returned positive.
Seven patients who underwent single-stage revision after
standard negative perioperative tests subsequently had multiple
positive intraoperative tissue cultures. The average IL‐6 level in
this group was 1400 pg/mL, and the IL‐6 level was considered
elevated in five of the seven patients. These five patients had
been followed for one year or less (average, 7.4 months) at the
time of this study. All five patients were treated with IV antibiotics postoperatively (average of 4.5 weeks), with four continuing with a period of oral antibiotic therapy (average, twelve
weeks). Two of the five patients were doing well when last seen,
S Y N O V I A L F L U I D I N T E R L E U K I N -6 A S A P R E D I C T O R
PER IPROSTHETIC SHOULDER INFECTION
OF
at four and six months, respectively. The other three patients
remained on chronic oral antibiotics, with persistent pain and
stiffness in the shoulder, although levels of serum-fluid markers
remained normal. Revision surgery was being considered for
two of these patients—one followed for one year and the other,
for seven months—because of the persistent symptoms. The
two patients with negative perioperative tests and negative IL-6
values who subsequently had positive cultures were treated
with IV antibiotics postoperatively (average of 5.0 weeks),
without additional oral antibiotic therapy. One patient had
persistent pain of unclear etiology at ten months postoperatively, and the other was doing well at one year postoperatively.
Both preoperative and intraoperative synovial fluid IL-6
levels were available for nine patients. IL-6 levels measured in
synovial fluid obtained intraoperatively demonstrated a moderate correlation with those in samples obtained from the same
patients preoperatively (Kendall correlation coefficient, 0.61;
p = 0.025) (Table III). However, only one of the nine patients
had conflicting preoperative and intraoperative IL-6 results
with regard to whether they were positive or negative based on
our cutoff value of 359.3 pg/mL. Synovial fluid IL-6 levels also
correlated both with the number of positive intraoperative
cultures (correlation coefficient, 0.58; p < 0.001) and the percentage of positive cultures based on the total number of cultures
obtained per case (correlation coefficient, 0.52; p < 0.001), but
they demonstrated a weak correlation with serum CRP levels and
no correlation with serum ESR (Table III). The percentage of
patients with elevated serum CRP levels was higher in the infection group (31% [four of thirteen] versus 0% [of eighteen] in
the no-infection group; p = 0.03), but the percentage of patients
with elevated ESR did not differ significantly between the infection and no-infection groups (38% [five of thirteen] versus 11%
[two of eighteen]; p = 0.1).
TABLE IV Comparisons of IL-6 Levels Between Categories
IL-6 Level (pg/mL)
Factor
Median
25th; 95th Percentile
191.6
24,621
77.5; 359.3
13,411.3; 33,498.5
Frozen-section histologic analysis
No acute inflammation
Acute inflammation
<0.001
Isolated microorganism
None
P. acnes
0.011
123.35
8531
71.2; 265.3
912.8; 33,439.7
Infection category
No infection
Probable infection
<0.001
93.25
1400.2
28.3; 232.9
400.9; 13,411.3
No-infection group
0.099
Arthroscopic rotator cuff repair
40.6
Revision shoulder arthroplasty
123.35
*Wilcoxon rank-sum test.
P Value*
25; 143.8
71.2; 265.3
68
TH E JO U R NA L O F B O N E & JO I N T SU RG E RY J B J S . O RG
V O L U M E 97-A N U M B E R 1 J A N UA RY 7, 2 015
d
d
d
Synovial fluid IL-6 levels also showed strong associations
with findings on frozen-section histologic analysis and growth of
P. acnes on culture (Table IV). The median IL-6 value in patients
with positive frozen sections was 24,621 pg/mL compared with
192 pg/mL in patients with negative frozen sections (p < 0.001).
Synovial fluid IL-6 levels were also elevated in the presence of a
P. acnes-positive culture (median, 8531 pg/mL). Of the seven
patients with positive intraoperative cultures for P. acnes, four had
frozen-section results that met the criteria for acute inflammation. Patients who met the criteria for probable infection (one
positive culture and other signs of infection, or no signs of infection but more than one positive culture) had higher IL-6 values
than patients in the no-infection group (median, 1400 versus 93
pg/mL; p < 0.001). Within the no-infection group, synovial fluid
IL-6 levels did not differ significantly between the patients who
underwent revision shoulder arthroplasty and those treated with
arthroscopic rotator cuff repair (p = 0.1; Table IV).
Discussion
JI is a major complication after shoulder arthroplasty and
remains a diagnostic challenge because of the subtle, often
nonspecific clinical presentation of indolent organisms such as
P. acnes. Early and successful identification of infection is critical
for determining subsequent medical and surgical management,
but current perioperative testing has been shown to be ineffective
in diagnosis. The aim of this study was to define the diagnostic
utility of synovial fluid IL-6 levels as a potential marker for PJI
of the shoulder. To our knowledge, this is the first study in the
literature to evaluate synovial fluid IL-6 as a marker of infection
after shoulder arthroplasty. Our results suggest that synovial fluid
IL-6 analysis is both sensitive and specific for predicting the presence of PJI in patients undergoing revision shoulder arthroplasty.
Recent efforts to identify more accurate markers of PJI
have targeted synovial fluid. Because serum markers may be
confounded by a concomitant acute or chronic inflammatory
state, assessments of local markers in the affected joints may
be better diagnostic tests. Deirmengian et al. identified several
synovial fluid biomarkers, including IL-6, whose levels were
substantially elevated in patients with hip or knee PJI10. In their
study, synovial fluid IL-6 analysis had a sensitivity and specificity
of 100% at a cutoff value of 13,350 pg/mL. Similarly, in a study
of several biomarkers, Jacovides et al. found synovial fluid IL-6
levels to be most strongly linked to hip and knee PJI, with 87%
sensitivity, and 100% specificity22. More recently, Gollwitzer
et al. assessed the diagnostic efficacy of synovial fluid IL-6
analysis in identifying staphylococcal hip and knee PJI using
cytometric bead arrays23. They reported an area under the curve
of 0.807, a sensitivity of 60%, and a specificity of 95% using a cutoff
value of 1896.6 pg/mL. Because of differences in the bioassays and
in the infecting organisms, it is difficult to compare our data
with the previously reported hip and knee literature. However,
using a cutoff of 359 pg/mL, we showed a sensitivity (87%) and
specificity (90%) similar to those in previously reported studies.
It must be noted that the high positive likelihood ratio of 8.45
should be interpreted as an increase in the pretest odds of having
an infection rather than the risk of having an infection. This
P
S Y N O V I A L F L U I D I N T E R L E U K I N -6 A S A P R E D I C T O R
PER IPROSTHETIC SHOULDER INFECTION
OF
distinction is particularly important for patients with a high
probability of having an infection, such as those presenting
with pain at the site of a shoulder arthroplasty. For these patients,
the odds tend to overestimate the risk.
The findings of this study demonstrate the clinical utility
and relevance of incorporating synovial fluid IL-6 analysis in
the diagnostic workup for patients undergoing revision shoulder
arthroplasty. When synovial fluid can be obtained, we recommend the use of an IL-6 assay as part of the preoperative evaluation. Eight of the nine patients from whom both preoperative
and intraoperative samples were obtained had agreement between the IL-6 results of the two assays. Ten patients who had
negative ESR and CRP values had two or more positive intraoperative cultures (probable-infection subgroup) and were
treated postoperatively with intravenous antibiotics. Only three of
these ten patients underwent a two-stage procedure, which is the
standard treatment when PJI is identified preoperatively or
intraoperatively. Of the remaining seven patients, who underwent a single-stage revision, five had intraoperative synovial
fluid IL-6 levels that were considered elevated. Also, the overall
median value was 1400.2 pg/mL in the probable-infection subgroup compared with 93.3 pg/mL in the no-infection group (p <
0.001) (Table IV). If the IL-6 data had been available preoperatively or intraoperatively for the five patients who had a singlestage revision and subsequently had positive cultures, a two-stage
exchange might have been chosen instead. This is important
particularly in light of the postoperative finding that three of the
five patients continued to have shoulder pain and stiffness with
persistent infection a potential concern. This not only demonstrates the potential efficacy of synovial fluid IL-6 analysis,
but also reiterates the insufficiency of the current diagnostic
modalities. Furthermore, synovial fluid IL-6 levels were significantly higher in patients who had growth of P. acnes on culture
(median, 8531 pg/mL) than in patients for whom cultures were
negative for infection (median,123.4 pg/mL, p = 0.011), indicating the potential to better detect infections caused by indolent
infecting organisms preoperatively or intraoperatively.
While the correlation between preoperative and intraoperative synovial fluid IL-6 values was only moderate in the nine
patients who had both assays, eight of the nine had preoperative
and intraoperative samples that were consistent in terms of
whether they were positive or negative according to the cutoff
value of 359.3 pg/mL. These are promising results in favor of
using synovial fluid IL-6 analysis as a preoperative test, but it is
difficult to make definitive conclusions on the basis of only nine
cases. The poor correlation of synovial fluid IL-6 analysis with
serum ESR and CRP levels is consistent with the poor accuracy of
the latter tests for diagnosing PJI of the shoulder8. The strong
association between a finding of acute inflammation on frozensection histologic analysis and elevated synovial fluid IL-6 levels is
reasonable considering that IL-6 is a pro-inflammatory marker.
The use of the MSD platform for synovial fluid analysis
provided some distinct advantages. First, the low-end analytical
sensitivity of the assay allowed detection of IL-6 in relatively
small volumes of synovial fluid, which enabled testing of a high
percentage of the enrolled patients. This will be an important
69
TH E JO U R NA L O F B O N E & JO I N T SU RG E RY J B J S . O RG
V O L U M E 97-A N U M B E R 1 J A N UA RY 7, 2 015
d
d
d
factor in the clinical setting, in which the ability to aspirate
larger volumes of synovial fluid from the shoulder can be
challenging. This is particularly true in the preoperative setting,
as in-office aspiration may yield scant or no fluid. In the current
study, a preoperative sample was obtained from only twenty-two
(48%) of the forty-six originally enrolled patients. Image-guided
aspiration may be the optimal method for obtaining a preoperative sample, but the volume yield may still be low. However,
demonstrating the efficacy of this biomarker is the first step in
developing a point-of-care test that can provide results intraoperatively, when a synovial fluid sample is easier to obtain (a 91%
success rate in the current study) and in cases in which preoperative fluid is not available. A second benefit of the MSD platform is that the plate-based nature of the technology is less prone
to the matrix issues with synovial fluid that influence bead-based
assays24. This led to a very low rate of analytical failures in our
study. Finally, the wide dynamic range of the assay allowed
changes of as much as 1000-fold over baseline to be readily
detected with a minimum of dilutions. This should allow better
differentiation of patients with elevated levels and may be helpful
in following responses to treatment in future studies.
There were some limitations to the current study. First,
there is no gold standard for the diagnosis of PJI following
shoulder arthroplasty. Although several tests and combinations of
clinical criteria have been used to classify PJI, it remains difficult to
diagnose with absolute certainty. Our use of different criteria for
PJI could have altered the primary results of our study, but this is
an inherent limitation of any study evaluating a diagnostic test for
infection. We chose to use a range of infection categories to better
stratify the patient population in terms of infection odds. This
allowed identification of one of the two patients with a falsenegative synovial fluid IL-6 analysis as having a probable infection.
A second limitation of our study was the small sample size of
patients with an infection. A large multicenter study is necessary to
further characterize infected patients. Finally, indolent infections
such as P. acnes tend to form biofilms and small foci surrounding
the implant that could elude the scant synovial fluid that is usually
present, making false-negative results an inevitable consequence
of synovial fluid testing in the shoulder.
Diagnosis of infection in patients undergoing revision
shoulder arthroplasty remains a diagnostic challenge due to
S Y N O V I A L F L U I D I N T E R L E U K I N -6 A S A P R E D I C T O R
PER IPROSTHETIC SHOULDER INFECTION
OF
the low virulence of the most common infecting organisms.
Current guidelines for diagnosis of PJI are not sensitive for diagnosing infection in such patients. This study demonstrates the
potential clinical benefit of performing synovial fluid IL-6 analysis preoperatively to detect infections at the sites of shoulder
prostheses. In addition, synovial fluid IL-6 analysis may be
helpful in the postoperative evaluation and treatment of cases
with unanticipated positive cultures. The diagnostic accuracy
of synovial fluid IL-6 analysis for identifying PJI may lead to
improved decision-making regarding management of shoulder
PJI, such as helping to determine the appropriate indications
for one or two-stage revisions.
Appendix
Descriptions of the synovial fluid and tissue handling,
MSD assay, and statistical methods are available with the
online version of this article as a data supplement at jbjs.org. n
Salvatore J. Frangiamore, MD
Anas Saleh, MD
Mario Farias Kovac, MD
Xiaochun Zhang, MD
Thomas M. Daly, MD
Eric T. Ricchetti, MD
Joseph P. Iannotti, MD, PhD
Departments of Orthopaedic Surgery (S.J.F., A.S., M.F.K., E.T.R., and J.P.I.)
and Clinical Pathology (X.Z. and T.M.D.),
Cleveland Clinic Foundation,
9500 Euclid Avenue,
A41, Cleveland, OH 44195.
E-mail address for J.P. Iannotti: [email protected]
Matthew J. Grosso, BS
Cleveland Clinic Lerner College of Medicine,
9500 Euclid Avenue,
Cleveland, OH 44195
Thomas W. Bauer, MD
Pathology and Laboratory Medicine Institute,
Cleveland Clinic Foundation,
9500 Euclid Avenue,
Cleveland, OH 44195
References
1. Coste JS, Reig S, Trojani C, Berg M, Walch G, Boileau P. The management of
infection in arthroplasty of the shoulder. J Bone Joint Surg Br. 2004 Jan;86(1):
65-9.
2. Kelly JD 2nd, Hobgood ER. Positive culture rate in revision shoulder arthroplasty.
Clin Orthop Relat Res. 2009 Sep;467(9):2343-8. Epub 2009 May 12.
3. Piper KE, Jacobson MJ, Cofield RH, Sperling JW, Sanchez-Sotelo J, Osmon DR,
McDowell A, Patrick S, Steckelberg JM, Mandrekar JN, Fernandez Sampedro M,
Patel R. Microbiologic diagnosis of prosthetic shoulder infection by use of
implant sonication. J Clin Microbiol. 2009 Jun;47(6):1878-84. Epub 2009 Mar 04.
4. Kobayashi H, Oethinger M, Tuohy MJ, Procop GW, Bauer TW. Improved detection
of biofilm-formative bacteria by vortexing and sonication: a pilot study. Clin Orthop
Relat Res. 2009 May;467(5):1360-4. Epub 2008 Nov 07.
5. Bauer TW, Parvizi J, Kobayashi N, Krebs V. Diagnosis of periprosthetic infection.
J Bone Joint Surg Am. 2006 Apr;88(4):869-82.
6. Strickland JP, Sperling JW, Cofield RH. The results of two-stage re-implantation
for infected shoulder replacement. J Bone Joint Surg Br. 2008 Apr;90(4):460-5.
7. Levy PY, Fenollar F, Stein A, Borrione F, Cohen E, Lebail B, Raoult D. Propionibacterium
acnes postoperative shoulder arthritis: an emerging clinical entity. Clin Infect Dis. 2008
Jun 15;46(12):1884-6.
8. Piper KE, Fernandez-Sampedro M, Steckelberg KE, Mandrekar JN, Karau MJ,
Steckelberg JM, Berbari EF, Osmon DR, Hanssen AD, Lewallen DG, Cofield RH,
Sperling JW, Sanchez-Sotelo J, Huddleston PM, Dekutoski MB, Yaszemski M, Currier
B, Patel R. C-reactive protein, erythrocyte sedimentation rate and orthopedic implant
infection. PLoS ONE. 2010;5(2):e9358. Epub 2010 Feb 22.
9. Alijanipour P, Bakhshi H, Parvizi J. Diagnosis of periprosthetic joint infection: the
threshold for serological markers. Clin Orthop Relat Res. 2013 Oct;471(10):3186-95.
10. Deirmengian C, Hallab N, Tarabishy A, Della Valle C, Jacobs JJ, Lonner J, Booth
RE Jr. Synovial fluid biomarkers for periprosthetic infection. Clin Orthop Relat Res.
2010 Aug;468(8):2017-23.
11. Di Cesare PE, Chang E, Preston CF, Liu CJ. Serum interleukin-6 as a marker of
periprosthetic infection following total hip and knee arthroplasty. J Bone Joint Surg
Am. 2005 Sep;87(9):1921-7.
70
TH E JO U R NA L O F B O N E & JO I N T SU RG E RY J B J S . O RG
V O L U M E 97-A N U M B E R 1 J A N UA RY 7, 2 015
d
d
d
12. Bottner F, Wegner A, Winkelmann W, Becker K, Erren M, Götze C. Interleukin-6,
procalcitonin and TNF-alpha: markers of peri-prosthetic infection following total joint
replacement. J Bone Joint Surg Br. 2007 Jan;89(1):94-9.
13. Coffey MJ, Ely EE, Crosby LA. Treatment of glenohumeral sepsis with a commercially produced antibiotic-impregnated cement spacer. J Shoulder Elbow Surg.
2010 Sep;19(6):868-73. Epub 2010 Apr 14.
14. Grosso MJ, Frangiamore SJ, Saleh A, Kovac MF, Hayashi R, Ricchetti ET,
Bauer TW, Iannotti JP. Poor utility of serum interleukin-6 levels to predict indolent
periprosthetic shoulder infections. J Shoulder Elbow Surg. 2014 Apr 9. Epub 2014 Apr 9.
15. Villacis D, Merriman JA, Yalamanchili R, Omid R, Itamura J, Rick Hatch GF 3rd.
Serum interleukin-6 as a marker of periprosthetic shoulder infection. J Bone Joint
Surg Am. 2014 Jan 1;96(1):41-5.
16. Kokebie R, Aggarwal R, Lidder S, Hakimiyan AA, Rueger DC, Block JA,
Chubinskaya S. The role of synovial fluid markers of catabolism and anabolism in
osteoarthritis, rheumatoid arthritis and asymptomatic organ donors. Arthritis
Res Ther. 2011;13(2):R50. Epub 2011 Mar 24.
17. Tsuchida AI, Beekhuizen M, Rutgers M, van Osch GJ, Bekkers JE, Bot AG,
Geurts B, Dhert WJ, Saris DB, Creemers LB. Interleukin-6 is elevated in synovial
fluid of patients with focal cartilage defects and stimulates cartilage matrix production in an in vitro regeneration model. Arthritis Res Ther. 2012;14(6):R262.
Epub 2012 Dec 03.
18. Trampuz A, Osmon DR, Hanssen AD, Steckelberg JM, Patel R. Molecular and
antibiofilm approaches to prosthetic joint infection. Clin Orthop Relat Res. 2003
Sep;(414):69-88.
S Y N O V I A L F L U I D I N T E R L E U K I N -6 A S A P R E D I C T O R
PER IPROSTHETIC SHOULDER INFECTION
OF
19. Dodson CC, Craig EV, Cordasco FA, Dines DM, Dines JS, Dicarlo E, Brause BD,
Warren RF. Propionibacterium acnes infection after shoulder arthroplasty: a diagnostic challenge. J Shoulder Elbow Surg. 2010 Mar;19(2):303-7. Epub 2009
Nov 01.
20. Zeller V, Ghorbani A, Strady C, Leonard P, Mamoudy P, Desplaces N.
Propionibacterium acnes: an agent of prosthetic joint infection and colonization.
J Infect. 2007 Aug;55(2):119-24. Epub 2007 Apr 05.
21. Workgroup Convened by the Musculoskeletal Infection Society. New definition
for periprosthetic joint infection. J Arthroplasty. 2011 Dec;26(8):1136-8.
22. Jacovides CL, Parvizi J, Adeli B, Jung KA. Molecular markers for diagnosis of
periprosthetic joint infection. J Arthroplasty. 2011 Sep;26(6)(Suppl):99-103.e1.
Epub 2011 May 13.
23. Gollwitzer H, Dombrowski Y, Prodinger PM, Peric M, Summer B, Hapfelmeier A,
Saldamli B, Pankow F, von Eisenhart-Rothe R, Imhoff AB, Schauber J, Thomas P,
Burgkart R, Banke IJ. Antimicrobial peptides and proinflammatory cytokines in
periprosthetic joint infection. J Bone Joint Surg Am. 2013 Apr 3;95(7):644-51.
24. Jayadev C, Rout R, Price A, Hulley P, Mahoney D. Hyaluronidase treatment of
synovial fluid to improve assay precision for biomarker research using multiplex
immunoassay platforms. J Immunol Methods. 2012 Dec 14;386(1-2):22-30.
Epub 2012 Aug 28.
25. Dabitao D, Margolick JB, Lopez J, Bream JH. Multiplex measurement of
proinflammatory cytokines in human serum: comparison of the Meso Scale Discovery electrochemiluminescence assay and the Cytometric Bead Array. J Immunol
Methods. 2011 Sep 30;372(1-2):71-7. Epub 2011 Jul 18.