Introduction

The diagnosis of prosthetic joint infection (PJI) in the absence of draining sinus, positive cultures or gross purulence is challenging. For this reason, additional supporting evidence is often necessary, and multiple biomarkers are being studied for their accuracy [1]. The prevalence of infection has been reported to be 0.4% to 2%, 5.6% and 15.8% following primary total knee arthroplasty (TKA) [2, 3], revision TKA [2] and a two-stage revision TKA [4], respectively.

The recently proposed definition by the Musculoskeletal Infection Society (MSIS) includes multiple secondary criteria such as erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), synovial white blood cell counts (WBC) and synovial polymorph nuclear leukocyte (PMN) percentage, D-dimer, alpha defensin as well as histology (“Appendix 1”) [4]. This definition also has a scoring system to improve the probability of diagnosing the PJI, but the authors have cautioned about its usefulness in the presence of adverse local tissue reactions, slowly growing organisms as well as crystal arthropathies [5]. No single laboratory test can accurately detect infection before revision hip or knee arthroplasty [3]. Tunney et al. [6] studied immunofluorescence microscopy and PCR amplification of bacterial DNA in 120 patients undergoing revision total hip surgery and concluded that the incidence of prosthetic joint infection is grossly underestimated by current culture detection methods.

Serum interleukin-6 (IL-6) is a pro-inflammatory cytokine released in the blood stream, especially in acute or low-grade chronic inflammations, in response to specific microbial molecules that precedes the appearance of other cytokines in the circulation and has been proposed as another biomarker in the diagnosis of chronic, low-grade PJI [7,8,9,10,11,12]. However, limited data are available regarding its actual role in the diagnosis of PJI. The diagnostic value of serum IL-6 has not been previously compared with synovial fluid markers, and limited data are available on its value in detecting or ruling out a PJI. Prior studies have not determined the strength of the correlation between IL-6 and other routine blood and synovial fluid markers in identifying PJI. The literature lacks evidence supporting the best screening or confirmatory test [11, 13]. We believe that identifying an ideal initial screening test could be very helpful in accurately ruling out or detecting infection following a TKA and avoid unnecessary or repeated joint aspirations as well as unindicated or wrong procedures that would have a significant impact on morbidity, mortality, cost of care and patient-reported outcomes.

This study was undertaken to determine the overall diagnostic value of serum IL-6 levels in infected TKA. The study goals were to specify: (1) the optimal threshold concentration of serum IL-6 to differentiate infected from noninfected TKA; (2) sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and accuracy of serum IL-6 in infected TKA; and (3) the diagnostic utility of serum IL-6. The thought was to find out its usefulness as either an individual marker or combined with other blood and synovial fluid markers of PJI for optimal use as a screening or confirmatory test.

Materials and methods

From January 2017 to December 2019, we prospectively identified patients undergoing revision TKA at our institution. The study was approved by our institutional review board.

A cohort of 59 knees (52 patients) had serum IL-6 evaluated simultaneously with ESR, CRP, synovial fluid WBC and synovial fluid PMNs (Table 1).

Table 1 Patient demographics
Full size table

The PJI was defined based on major criteria outlined in the MSIS definition (“Appendix 1”) [4] as the literature lacks adequate supporting evidence for the accuracy of the individual minor criteria [5, 14, 15]. These patients presented with a painful TKA underwent treatment for infection or another implant-related problem and were managed by a single surgeon at a single institution. Patients were excluded from the study when either data were incomplete or if patients had active inflammatory diseases that could be responsible for a preexisting higher level of IL-6.

All samples were collected, analyzed and processed (Table 2) within 6 h of procurement at our institution’s laboratory. Samples that were clotted or grossly bloody were excluded from the study.

Table 2 Methods of testing the markers of prosthetic joint infection
Full size table

Based on the tests, results were plotted using receiver operating characteristic (ROC) curves (Fig. 1a, b) to determine the optimal threshold concentration of serum IL-6, CRP, ESR, synovial fluid WBC and synovial fluid PMN percentage in the PJI. Their sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), accuracy and likelihood ratios within the 95% confidence interval were determined. These markers were then studied individually as well as in combination using a first strategy where either of the markers in the mix could have laboratory values above the threshold concentration and second strategy where all laboratory values must be above the threshold concentration.

Fig. 1
figure 1

a Receiver operating characteristic curves for serum interleukin 6, ESR and CRP in infected total knee arthroplasty. b Receiver operating characteristic curves for serum interleukin 6, synovial fluid WBC and synovial fluid PMN percentage in infected total knee arthroplasty. The diagonal line indicates area under the curve of 50% or less and indicates the test is not useful in the diagnosis of infected total knee arthroplasty. The threshold value determined from the ROC curves was the value closest to 1 (upper left corner on graph) and is the most accurate value for that marker in the diagnosis of PJI

Full size image

Fischer’s exact tests were used to analyze the statistical differences in performance of each test in infected versus noninfected group, and p value of < 0.05 was considered to be statistically significant. Mann–Whitney U test and two-sample test were used to evaluate the univariate data. Spearman’s rank correlation was used to correlate and compare IL-6 with other markers in the MSIS definition of PJI [4].

Results

The optimal threshold concentration of serum IL-6 for differentiating between an infected versus noninfected TKA is 9.14 pg/ml as determined by the ROC curve (Fig. 1a). This value is almost three times higher than the average value for our laboratory (< 3 pg/ml).

When tested alone, an IL-6 threshold concentration of 9.14 pg/ml has 81.3% sensitivity (95% confidence interval CI, 54% to 95%), 63% specificity (95% CI, 43% to 80%), 85% NPV (95% CI, 61% to 96%), 56.5% PPV (95% CI, 35% to 76%) and 70% accuracy for diagnosing PJI (Table 3).

Table 3 Statistics of the individual and combined tests based on the threshold valuesa of the markers for infected total knee arthroplastyb
Full size table

The optimal threshold concentrations for ESR (> 27 mm/h), CRP (> 17 mg/L), synovial WBC (> 1804 cells/10−3 cm3), synovial PMNs (> 86%) from the ROC curves were determined, and the values were comparable to those published in the available literature (Table 4). The sensitivity of serum IL-6 is higher than that of ESR (70%), CRP (67%) and synovial WBC (67%) (Table 3; “Appendix 2”) and almost equivalent to that of the synovial PMN percentage (82%). The serum IL-6 concentration of 9.14 pg/ml has a highest overall accuracy of 70% when compared with ESR (67%), CRP (67%), but is less accurate than synovial fluid WBC (74%) and synovial fluid PMN percentage (78%) based on the study findings.

Table 4 Comparison of threshold value of markers of prosthetic joint infection with those from the published studies
Full size table

When serum IL-6 results are combined with those of ESR and CRP with either of these values being above the threshold concentration, the sensitivity of detecting PJI increases up to 94% (95% CI, 67% to 100%) and NPV increases up to 91% (95% CI, 57% to 100%). Similarly, if serum IL-6 results are combined with those of synovial PMNs or synovial WBC and if either of these values was above the threshold level, then the sensitivity, as well as NPV, increases up to 100% (95% CI, 73% to 100%). We also found out that if serum IL-6 results are combined with those of synovial WBC and when both values were above the cutoff level, the specificity and accuracy of detecting PJI were highest and went up to 90% (95% CI, 67% to 98%) and 78.1%, respectively (Table 3; “Appendix 2”).

Serum IL-6 levels strongly correlate with those of ESR (p < 0.0001), CRP (p < 0.0003), synovial fluid WBC (p < 0.0400) and synovial fluid PMN (p < 0.0070) when tested simultaneously in patients with infected TKA (Table 5).

Table 5 Strength of correlation of blood and synovial fluid markers in infected total knee arthroplasty when tested simultaneously prior to the definitive treatment
Full size table

The most common microorganisms obtained from patients with two or more positive cultures included methicillin-sensitive Staphylococcus aureus (28%), Streptococcus species (17%), Coagulase-negative staphylococci (14%), Propionibacterium acnes (14%), Corynebacterium (7.5%), methicillin-resistant Staphylococcus aureus (5%), Enterococci (5%), Candida albicans (5%) and others (4.5%).

Discussion

Accurate diagnosis of chronic, low-grade PJI remains elusive and least invasive, and workup recommended for the earliest possible way to screen a painful prosthetic joint for infection continues to be inflammatory blood markers like ESR and CRP, but they have reduced sensitivity and specificity [11]. The newer biomarkers like alpha defensin or D-dimer lack adequate multi-institutional, randomized controlled study data on their role in accurately screening or diagnosing the PJI, especially in the presence of comorbid conditions that could alter their levels in the blood and synovium [12, 16]. Our goal was to identify the most straightforward, best and most cost-effective initial screening tool from blood or synovial fluid markers to identify the low-grade prosthetic infections. IL-6, an acute phase protein, is one of the earliest pro-inflammatory cytokines released in the bloodstream by the injured tissue, and it not only acts as a precursor for CRP but also triggers a lot of other biomarkers responsible for neutrophil chemotaxis to the inflamed site [8].

According to our study, the optimal threshold concentration for serum IL-6 in the diagnosis of infected TKA is 9.14 pg/ml. This threshold value obtained in our study using ROC curves is close to that determined by two other studies and confirms the existing knowledge of the usefulness of IL-6 in the diagnosis of PJI [7,8,9]. Prior studies have determined that serum IL-6 could be diagnostic of PJI if the optimal value was above 10 pg/ml for infected THA &/TKA (based upon median of 17 patients) [11] or 12 pg/ml for infected THA (n = 69) [9, 10]. Data supporting our finding are more robust for infected TKAs. The normal range for serum IL-6 of 0 to 3 pg/ml in healthy patients was based on the high-sensitivity ELISA technique used with Luminex 100 analyzer at University Cytokine Reference Laboratory. The serum IL-6 level in normal individuals is approximately 1 pg/ml [10] with modest elevations of up to 10 pg/mL in patients with inflammatory diseases or malignancies [17]. The potential advantage of using this marker is that it is normal in patients with aseptic loosening [18]. It may show large elevations of 30 to 430 pg/mL for only up to 70 h after a routine arthroplasty, and then they return to normal [17, 19] and so the level of this marker is not accurate for diagnosis of prosthetic infection for up to 3 days after the surgery [19].

Serum IL-6 as an individual marker has the highest sensitivity and highest NPV (rules out PJI) but lower specificity and PPV for PJI, when compared with all other routine markers analyzed in this study. These findings partially concur with those from the study published by Di Cesare et al. [10] that the serum IL-6 has the highest sensitivity and highest NPV in the diagnosis of PJI, but our study contradicts their finding that it also has higher specificity and PPV. This contradiction is possibly exposed due to the larger sample size in our study.

Serum IL-6, when combined with synovial markers (WBC, PMN), demonstrates highest sensitivity (100%), specificity (90%), PPV (78%) as well as NPV (100%) among all combinations tested (“Appendix 2”). The significance of this finding is that this combination could be very reliably used to rule out 100% of infections, thus improving the screening of an occult PJI in a painful TKA. None of the studies published so far have demonstrated this finding.

The strengths of the study are that it was performed at a single institution, the markers were tested at an only cytokine reference laboratory and they were exclusively analyzed in a painful TKA.

The weakness of the study is that (1) a bigger sample size could have strengthened the findings in this study. (2) Patients with a different organism mix as regards low- or high-virulent organisms could impact the overall outcomes due to a variable level of inflammatory response launched by that particular low- or high-virulent organism (Table 6). Also, the major limitations of this study are that (3) we did not test other similar biomarkers and compare or correlate their utility as we were looking for the most sensitive or specific biomarker that is expressed in the bloodstream and try to find a perfect combination of biomarkers to identify PJI. (4) The criteria used to diagnose infection (positive culture on solid media or sinus tract communicating with the prosthesis) may have introduced bias as there is a possibility of identifying more patients with PJI if additional minor criterion from MSIS definition would have been used. But we decided to exclude the minor criteria as most of them are less specific than the major criteria and are still being tested for their efficacy and accuracy [20, 21]. (5) There is also a possibility of up to 20% variation in the laboratory value of interleukin 6 if it is tested in different laboratories, using different ELISA kits, different analyzers, calibration and manufacturer-recommended settings [17, 22].

Table 6 Sensitivities and specificities of markers of infected total knee arthroplasty due to inflammatory response generated by specific organism groups at the threshold values
Full size table

This study concludes that the optimal threshold concentration of serum IL-6 at 9.14 pg/ml is diagnostic of infected TKA. Serum IL-6 levels strongly correlate with all the tested markers of PJI and could be reliably used in combination with synovial fluid markers to improve the infection screening in a painful TKA.