It is very difficult to identify an early postoperative infection after total knee or hip arthroplasty because there is no an absolute diagnostic criteria. In this study, we determined interleukin-6(IL-6) in serum had the highest accuracy as a marker for diagnosing early PJI, followed by the NRL, the C-reactive protein, the erythrocyte sedimentation rate, and the white blood cell count.
Interleukin-6 is an inflammatory cytokine produced by activated monocytes and macrophages and induces to product some major acute-phase proteins, including CRP . Some studies have showed that the IL-6 in serum was a valuable marker for diagnosis of PJI. Di Cesare et al.  showed serum IL-6 was more accurate marker than either ESR or the CRP level for diagnosing chronic PJI. A meta-analysis  published that IL-6 in serum was a potential superior diagnostic test compared to the conventional CRP and ESR, with the pooled sensitivity and specificity of 97% and 91%, respectively. At present there were no reports about IL-6 in serum as a useful marker for diagnosis of early PJI. Maniar et al.  found the normal trajectory of IL-6 had a more rapid increase and quicker return to normal than CRP after uncomplicated TKA. Although the author could not comment specifically on the value in diagnosing PJI, any deviation of IL-6 from a known normal trajectory could promptly make a decision to perform knee aspiration to diagnose early PJI. However, in our study, when we identified the cutoff values for the IL-6 at 8.07 pg/ml, the sensitivity, specificity was 80%, 76.2%, respectively. IL-6 in serum may be a useful marker for diagnosing early PJI. This is the first to assess the use of serum IL-6 to diagnose early PJI.
The serum IL-6 was not often detected because of expensive costs. Interestingly, another indicator, NLR performed very well and had the high AUC (0.804). The accuracy of the value of NLR was less than IL-6 but greater than CRP. It may be more suitable for screening to diagnose early PJI because it is convenient to be obtained in daily practice without extra costs.
The value of the NRL had been shown a significant association with infection. Dogruel et al.  demonstrated there were statistically significant correlations between NLR and prolonged hospital stay and postoperative antibiotic doses for the treatment of odontogenic infection. Josse et al.  reported that the values of NLR ≥ 2.3 in the pre-operation may be an independent predictor for major operative complications such as wound infection in patients undergoing colorectal surgery. Bolat et al. investigated the relationship between NLR and early postoperative infection as a complication of penile prosthesis implantation (PPI). And the values of NLR may be a valid laboratory predictor for early postoperative infectious complications for patients who went through PPI. De Jager et al. described the NLR had the highest AUC of 0.73 and differed significantly from the CRP level of 0.62. The NLR was better value in predicting bacteremia than routine parameters like CRP. Yombi et al.  reported that the NLR had a distribution trajectory in a standard post-operative period after TKA, and returned to normal values faster than CRP. It was potentially a better biomarker to follow post-operative inflammation or early infection after TKA. On the basis of our data, with the calculated threshold of NLR set at 2.13, the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) was 85%, 68.3%, 34.7%, and 95.8%, respectively. To our knowledge, this is the first study to assess the NLR in diagnosing early PJI.
Although there were some literatures that elaborate possible explanations about the relation between elevated NLR and the development of postoperative infection [20–22], it remained unclear about the exact mechanism. But the most crucial finding of this study was that NLR may be a useful parameter for diagnosing early PJI.
There are some limitations of this study that should be considered. First, this is retrospective study and the inherent limitations exist. There may be some selection bias because several potential cases were excluded due to no record of the serum ESR, CRP, or IL-6. Although this study is a retrospective design, it may provide some useful information for diagnosing early PJI. Second, there is no consensus on definition of the early PJI. We defined 90 days as early PJI and shorter time needed to be detected. Finally, we recruited 121 cases to the study. Although the total sample size was relatively sufficient, there was a small sample size in infected cases that limited the statistical power of our conclusions. Therefore, our findings should be confirmed in a larger study and at multiple institutions.