Serum and Synovial Fluid Interleukin-6 for the Diagnosis of Periprosthetic Joint Infection

A gold standard for diagnosis of periprosthetic joint infection (PJI) has not yet been established. The objective of this study was to evaluate the diagnostic value of serum and synovial fluid interleukin (IL)-6 levels for PJI. The MEDLINE and EMBASE databases were searched for studies describing PJI diagnosis using serum and synovial fluid IL-6 and published between January 1990 and October 2016. Seventeen studies were included in the analysis. The pooled sensitivities of serum and synovial fluid IL-6 were 0.72 (95% confidence interval [CI]: 0.63–0.80) and 0.91 (95% CI: 0.82–0.96), respectively. The pooled specificities of serum and synovial fluid IL-6 were 0.89 (95% CI: 0.77–0.95) and 0.90 (95% CI: 0.84–0.95), respectively. The pooled diagnostic odds ratios (DORs) of serum and synovial fluid IL-6 were 20 (95% CI: 7–58) and 101 (95% CI: 28–358), respectively, and the pooled areas under the curve (AUCs) were 0.83 (95% CI: 0.79–0.86) and 0.96 (95% CI: 0.94–0.98), respectively. Synovial fluid IL-6 had high diagnostic value for PJI. Although serum IL-6 test was less sensitive than synovial fluid IL-6 test, it may be regularly prescribed for patients with prosthetic failure owing to its high specificity.

Seventeen articles were included in the current analysis (Table 1). Nine studies were conducted in the United States, three in Germany, and one study each in Sweden, Argentina, Austria, United Kingdom, and Egypt. Nine articles described the diagnostic accuracy of serum IL-6 for PJI, and six articles focused on synovial fluid IL-6 test. Two additional studies investigated both serum and synovial fluid IL-6 assessments for PJI diagnosis. All studies included in the current meta-analysis were prospective studies. Consecutive patient enrolment was used in four studies, while the others did not indicate the type of patient enrolment. Three studies included patients with shoulder arthroplasty, while the others focused on knee and/or hip arthroplasty. The number of participants ranged from 35 to 120, and the mean age ranged from 58 to 72 years. The cut-off values of serum and synovial fluid IL-6 levels for PJI detection in the selected studies ranged from 2.6 to 10.4 pg/mL and 359.3 to 13,350 pg/mL, respectively. All studies were evaluated using the QUADAS tool, and showed moderate to high quality. Diagnostic Accuracy. Pooled sensitivity, specificity, area under the curve (AUC), and diagnostic odds ratio (DOR) results are shown in Fig. 2. The pooled sensitivity and specificity for PJI diagnosis using serum IL-6 were 0.72 (95% CI: 0.63-0.80) and 0.89 (95% CI: 0.77-0.95), respectively. The pooled sensitivity and specificity for PJI diagnosis using synovial fluid IL-6 were 0.91 (95% CI: 0.82-0.96) and 0.90 (95% CI: 0.84-0.95), respectively. The pooled DORs for PJI diagnosis using serum and synovial fluid IL-6 were 20 (95% CI: 7-58) and 101 (95% CI: 28-358), respectively. The pooled AUCs for serum and synovial fluid IL-6 tests were 0.83 (95% CI: 0.79-0.86) and 0.96 (95% CI: 0.94-0.98), respectively. Heterogeneity was evaluated by I 2 . The I 2 of serum IL-6 was 81, indicating substantial variation among the included studies. No heterogeneity was found for synovial fluid IL-6 test (I 2 = 0).

Subgroup Analysis.
Results of the subgroup analysis are presented in Table 2. Both serum and synovial fluid IL-6 tests showed reliable diagnostic accuracy for PJI after hip and knee arthroplasty. The sensitivities of serum and synovial fluid IL-6 for hip and/or knee PJI diagnosis were 0.74 (95% CI: 0.63-0.80) and 0.92 (95% CI: 0.82-0.97), respectively. The specificities of serum and synovial fluid IL-6 for hip and/or knee PJI diagnosis were 0.92 (95% CI: 0.79-0.97) and 0.91 (95% CI: 0.83-0.95), respectively. The current analysis also showed that the cut-off value could influence the diagnostic accuracy of both serum and synovial fluid IL-6 tests. When the cut-off value was greater than 10 and 2,300 pg/mL for serum and synovial fluid IL-6 tests, respectively, the diagnostic accuracy was improved. The diagnostic accuracy of synovial fluid IL-6 was not affected by the presence of inflammatory

Discussion
The current meta-analysis indicated that synovial fluid IL-6 can be used for the diagnosis of PJI after joint arthroplasty. Although serum IL-6 is less sensitive than synovial fluid IL-6, it is one of the best serum biomarkers for PJI detection. Because of its reliable post-test probability, serum IL-6 assessment should be included as a regular test for patients with prosthetic failure. Treatment for PJI involves intensive debridement, one-or two-stage implantation, and long-term antimicrobial therapy; in contrast, a generic revision procedure is typically sufficient to treat aseptic loosening 6 . Thus, a precise diagnosis is necessary for surgeons to lay out the operation plan. However, PJI is difficult to diagnose before revision surgery in the absence of uniform criteria and a gold standard test. In the past decade, several meta-analyses have been performed to assess the diagnostic accuracy of various established tests for PJI detection (Table 3 and Fig. 5) 9, 10, 32-43 . Traditionally, periprosthetic tissue culture, which has a positive detection rate of 0.70-0.90, is regarded as the gold standard test for PJI diagnosis 8,[44][45][46] . Two prior meta-analyses have demonstrated that sonication and polymerase chain reaction can increase the diagnostic accuracy of prosthetic biopsy 41,42 . Unfortunately, these intraoperative tests cannot be performed before revision surgery.
Radiography is the most basic test prescribed for patients with joint pain after arthroplasty. Implant migration with transcortical sinus tracts is a reliable indicator for PJI 6 . However, septic hip prosthetic loosening typically shows normal radiographic findings 47 . Because metal-related artefacts are used in these patients, computed tomography and magnetic resonance imaging cannot be used for PJI diagnosis. Nuclear medicine appears to have a high sensitivity for PJI detection. The pooled sensitivity and specificity of leukocyte scintigraphy were 0.88 and 0.92, respectively, and those for fluorodeoxyglucose positron emission tomography were 0.86 and 0.93, respectively. The disadvantages of nuclear medicine for PJI diagnosis are high cost and radiation dose 8 .
Serum biomarkers are commonly utilised to detect infections worldwide. According to the guidelines specified by the Infectious Diseases Society of America (IDSA), the erythrocyte sedimentation rate and CRP should be evaluated for all patients suspected as having PJI 8 . However, serum biomarker levels can be easily influenced by antibiotic therapy, systemic infection, trauma, and surgery 48 ; thus, these biomarkers may indicate inflammation rather than infection. According to a previous study, procalcitonin (PCT) is the most specific serum biomarker of PJI with a specificity of 0.92 32 .
Recently, several studies have investigated the diagnostic capacity of serum and synovial fluid IL-6 for PJI. IL-6 is a 212-amino acid interleukin encoded by a single gene mapped to chromosome 7 in humans 49 . The correlation between high levels of IL-6 in body fluids and local acute bacterial infection was reported in 1989 50 . The level of serum IL-6 is typically less than 10 pg/mL in healthy adults 51 . IL-6 is an anti-inflammatory cytokine and a mediator of infection response 52 . It is considered the major mediator of acute phase protein production by hepatocytes 53 . Serum CRP level increases after 4-6 h of IL-6 stimulation 54 , indicating that IL-6 is a relatively sensitive biomarker of early-stage immune response. As reported in a previous meta-analysis that assessed the findings of three studies, the pooled sensitivity and specificity of serum IL-6 were 0.97 and 0.91, respectively for PJI detection 9 . This remarkable diagnostic accuracy is inconsistent with the evidence presented in the current report. In the present meta-analysis, 11 studies that assessed serum IL-6 levels from 2005 to 2015 were included. The present study demonstrated that low detection sensitivity (0.72) was a major limiting factor of using serum IL-6 for PJI diagnosis. Notably, two prior studies that included 103 patients reported poor sensitivity (0.13-0.14) for serum IL-6 to detect periprosthetic shoulder infections 26,28 . The poor sensitivity reported in these studies may be related to the high proportion of low-grade infections in shoulder PJI. As has been established previously, different microorganisms cause PJI of different joints 6 . Low-virulent bacteria such as Propionibacterium acnes are frequently detected in patients with periprosthetic shoulder infection 55 . Nevertheless, Ettinger et al. reported a promising result for serum IL-6 testing to predict low-grade PJI with a sensitivity of 0.80 29 . Additional large-scale studies are necessary to determine the diagnostic value of serum IL-6 for shoulder PJI.
Joint aspiration is conventionally performed in patients with suspected PJI. Synovial fluid culture is an accurate diagnosis method with a sensitivity and specificity of 0.72 and 0.95, respectively 33 . Thus, results of aspiration culture can be used to guide treatment plans. In recent years, synovial fluid biomarkers have received significant attention for their diagnostic value for PJI detection. The pooled sensitivities of synovial fluid WCC, polymorphonuclear leukocytes, CRP, α-defensin, and leukocyte esterase are 0.88, 0.90, 0.92, 1.00, and 0.81, respectively, and their pooled specificities are 0.93, 0.88, 0.90, 0.96, and 0.97, respectively 32, 34-36 . Synovial fluid α-defensin appears to be the most valuable test to diagnose PJI (Fig. 5); however, it is not a universal test and is expensive (760 USD per test) 36 . The use of the α-defensin test for PJI diagnosis was first reported in 2014 56,57 . To our knowledge, it is not yet used in clinical practice in China. The synovial fluid IL-6 test is considerably more common and economical than the synovial fluid α-defensin test. The present meta-analysis indicated that abnormally high synovial fluid IL-6 level is a strong indicator of PJI, with an AUC of 0.96. Notably, our subgroup analysis indicated the high diagnostic value of the synovial fluid IL-6 test was not influenced by the presence of inflammatory diseases, which renders the test more advantageous for PJI detection in clinical practice. However, as with the serum IL-6 test, the optimal cut-off value of synovial fluid IL-6 is yet to be determined (range: 359.3-13,350 pg/mL).
The current meta-analysis has several strengths: 1) to the best of our knowledge, this is the first meta-analysis to report the diagnostic capacity of synovial fluid IL-6 for PJI detection; and 2) when compared to a prior meta-analysis, the current analysis included more studies (11 articles) assessing serum IL-6 test for PJI diagnosis, and showed that serum IL-6 was less sensitive than synovial fluid IL-6. We believe that this meta-analysis may aid orthopaedists to improve the diagnostic accuracy for PJI in clinical practice.
However, this meta-analysis has several limitations. First, the result of statistical analysis indicated significant heterogeneity for the serum IL-6 test. Two studies reported exceedingly low diagnostic values for the serum IL-6 test in the detection of periprosthetic shoulder infections 26,28 . The heterogeneity was still present after the two studies on shoulder PJI were excluded (I 2 = 79), which lowered the reliability of the findings. Second, a gold standard test for PJI diagnosis could not be proposed because multiple reference standards were used in the studies included in this meta-analysis. Third, antibiotic therapies and systemic inflammatory diseases may affect the diagnostic accuracy of serum IL-6. Only few studies excluded patients with prior antibiotic therapy or other infections. Fourth, the current study only included articles published between January 1990 and October 2016. Lastly, we did not register a protocol for the current study in PROSPERO.   The results of this meta-analysis showed that synovial fluid IL-6 has significant diagnostic value for PJI diagnosis, particularly in periprosthetic knee and hip infections. Although serum IL-6 test is less sensitive than synovial fluid IL-6 test, it can be regularly prescribed for patients with prosthetic failure owing to its high specificity.

Methods
The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement was used to guide methods for the current study 58 .
Search Strategy. The MEDLINE and EMBASE databases were searched for studies that utilised the serum or synovial fluid IL-6 test to diagnose PJI between January 1990 and October 2016. The search was based on the following terms: interleukin, IL-6, inflammatory, arthroplasty or replacement, serum or synovial fluid, sensitivity or specificity, septic or septically, prosthesis infection, infectious or infected, and diagnose or diagnostic. Additional entries were identified through the bibliographies of the selected studies and reviews.
The following inclusion criteria were used to evaluate the articles: (1) articles indicated the date of serum or synovial fluid IL-6 assessment of patients undergoing revision arthroplasty; (2) the diagnosis of PJI was made based on the IDSA guideline, Musculoskeletal Infection Society (MSIS) criteria, or American Academy of Orthopaedic Surgeons (AAOS) guideline 8, 45, 46 ; (3) sufficient data were described for the calculation of the true-positive (tp), false-negative (fn), false-positive (fp), and true-negative (tn) values; (4) the number of participants were more than 10; and (5) the article was written in English.
When the same patients were included in multiple articles, findings from the most detailed study were used. When an article described more than one cut-off value, the most efficient value was used in the present analysis. After a full-text assessment, article quality was assessed using the QUADAS tool 59 . Two independent reviewers performed the evaluation, and a third reviewer was solicited to resolve disagreements. Data Extraction. Two reviewers independently completed data extraction from eligible studies. The characteristics of each study were recorded in a standardised form to include the following details: name of the first author, publication year, number of patients, mean age of patients, study design, type of patient enrolment, location of the operation, cut-off value, and reference standards. A third reviewer was solicited to check for discrepancies and resolve conflicting findings of the first two reviewers. Statistical Analysis. Statistical analysis was performed by two independent reviewers. The tp, fn, fp, and tn values were calculated according to the statistical data extracted from eligible articles. Quantitative indicators, including sensitivity, specificity, AUC, and DOR, were calculated to estimate the diagnostic value of the IL-6 test for PJI detection. The PLR, NLR, and post-test probability were used to calculate the clinical utility of the IL-6 test. I 2 was calculated to evaluate study heterogeneity 60 Table 3. Diagnostic value of different diagnostic method for the diagnosis of PJI. If one diagnostic method was reported by more than one meta-analyses, the most detailed and/or recent one was included in Table 3. ESR: Erythrocyte sedimentation rate; FDG-PET: 18F-fluoro-2-deoxyglucose positron emission tomography. PCR: Polymerase chain reaction.
random effects model 61 . Subgroup analysis was performed to determine the effect of various characteristics on the diagnostic accuracy of the IL-6 test for PJI detection. Publication bias was evaluated using Deeks' funnel plot asymmetry test 62 . STATA version 14 (StataCorp, College Station, TX, USA) was used for other statistical analyses.
Data Availability. All data generated or analysed during this study are included in this published article.