Total joint replacement (TJA) is the most effective treatment for advanced arthritis, but periprosthetic joint infection (PJI) is a serious complication and a major reason for postoperative revision (1, 2). The incidence of PJI after arthroplasty was 0.7% (3). A study has shown that each patient suffered from PJI would pay at least $15,000–$30,000 (4). Therefore, early diagnosis of PJI and effective intervention are very to improve the prognosis after total joint replacement. A large number of clinical studies have shown that early diagnosis can not only preserve the prosthesis to a certain extent, but also the infection control rate can reach 70% (5). Currently, there is no “gold standard” for early diagnosing PJI. The diagnosis of periprosthetic joint infection (PJI) is a challenging problem in the clinic. This leads to the delayed diagnosis and treatment of PJI. As a result, this catastrophic complication can seriously diminish patient quality of life and increase the financial burden on families and societies (2).

Serum samples are readily available and are especially important for patients without Synovial fluid samples. As we know, many serum markers have been used to diagnose PJI early. Erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) are recommended as essential indicators for the diagnosis of PJI by the American Academy of Orthopaedic Surgeons (AAOS) and the Musculoskeletal Infection Society (MSIS) guidelines due to their superior sensitivity and specificity (6, 7). Zhang et al. first reported that serum platelet (PLT) is a promising marker for the diagnosis of deep surgical site infection after open induction internal fixation for traumatic limb fractures (8). Klim, S.M. et al. showed that Fibrinogen (FIB) is a cost-efficient and practical marker for diagnosing PJI (9, 10). Procalcitonin (PCT) is commonly used for the diagnosis of systemic infection (11–13). However, its diagnostic value for PJI is controversial, and there is no universal threshold value (14–16). To further clarify its diagnostic value, this study evaluated the value of PCT in the diagnosis of PJI by comparing it with CRP, ESR, PLT, and FIB.

In this retrospective study, we sought to: (1) the performance of PCT in distinguishing chronic PJI and AL (Aseptic Loosening) by comparing with other inflammation indicators; (2) the value of PCT combined with CRP, ESR, PLT, or FIB for diagnosing PJI.

A total of 77 patients were included for the final analysis. According to the MSIS criteria, 41 patients were identified as the periprosthetic joint infection (PJI) group (27 hips and 14 knees). In comparison, 36 patients were identified as the aseptic loosening (AL) group (33 hips and 3 knees). The mean age in the PJI group was 64.5 ± 10.0 years; of them, 17 were men and 24 were women. The mean age in the AL group was 65.9 ± 9.8 years; of them, 11 were men, and 25 were women. The two cohorts did not differ statistically in age (p = 0.529) and gender (p = 0.321). At the same time, there were no statistically significant differences between groups with diabetes mellitus (p = 0.094) or hypertension (p = 0.683). However, a statistical significance was shown in joint type (p = 0.014), time from primary arthroplasty to the first reoperation (p = 0.001), and time from symptom onset to the first reoperation (p = 0.001) between the two groups. The characteristics of the recruited patients were depicted in Table 1.

We evaluated the tested markers (CRP, ESR, PLT, FIB, and PCT) for all included patients. All patients in the PJI group had significantly higher values for the four markers (CRP, ESR, PLT, and FIB) compared with the AL group (all P < 0.05). Unfortunately, there was no significant difference for PCT between the two groups (P = 0.747). The details were shown in Table 3. We also listed the normal ranges for these tested markers in Table 3. As shown in Table 4 for culture organisms in the PJI group, Staphylococcus aureus and Streptococcus agalactiae were the two most common pathogens. We classified these pathogens into two groups (Table 5). There were no significant differences in all tested markers (CRP, ESR, PLT, FIB, and PCT) between the two groups (P > 0.05). That is to say, the results of the tested markers had nothing to do with the species of pathogens.

All tested markers (CRP, ESR, PLT, FIB, and PCT) were evaluated and plotted in the ROC curves (Figure 2). The AUCs for CRP, ESR, PLT, FIB, and PCT were 0.845 (95% CI 0.755–0.936, p < 0.001), 0.817 (95% CI 0.718–0.916, p < 0.001), 0.728 (95% CI 0.613–0.843, p < 0.001), 0.810 (95% CI 0.710–0.910, p < 0.001) and 0.504 (95% CI 0.373–0.635, p = 0.950), respectively (Table 6). The ROC curves showed that CRP had the highest AUC, followed by ESR, FIB, PLT, and PCT. The AUCs of CRP, ESR, and FIB ranged from 0.800–0.899, which demonstrates a good diagnostic value for PJI. The AUC of FIB was between 0.7 and 0.8, indicating a fair diagnostic value for PJI. In contrast, PCT had an AUC of 0.504 (lower than 0.6), the lowest value, indicating an inferior diagnostic value for PJI. Further analyses of the diagnostic value of PCT combined with other markers for PJI were conducted in order to improve their diagnostic accuracies. Higher AUC values indicating better diagnostic accuracy were obtained for the combinations of PCT and CRP (AUC = 0.870) (95% CI, 0.774–0.936), PCT and ESR (AUC = 0.817) (95% CI, 0.712–0.896), PCT and PLT (AUC = 0.731) (95% CI, 0.617–0.825), PCT and FIB (AUC = 0.815) (95% CI, 0.710–0.894). Among them, the combined analysis of PCT with CRP had the highest AUC. In conclusion, combining serum PCT with one of the other markers can improve their diagnostic accuracies (Table 6).

The serum PCT indicated a sensitivity of 19.51% and a specificity of 83.33% for diagnosing PJI. When the optimal cut-off value for PCT was set as 0.05 ng/ml, its positive likelihood ratio (PPV) and negative likelihood ratio (NPV) were 57.1% and 47.6%, respectively. Based on the ROC analysis, when CRP was above 9.75 mg/L, the sensitivity, specificity, PPV, and NPV were 87.80%, 72.22%, 78.3%, and 83.9%, respectively. Using a cut-off value for ESR of 45 mm/h, the sensitivity, specificity, PPV, and NPV were 67.50%, 91.67%, 90.0%, and 71.7%. Using a cut-off value for PLT at 291 × 10⁹/L, the sensitivity, specificity, PPV, and NPV were 63.41%, 80.56%, 78.8%, and 65.9%, respectively. Using Youden's index, the optimal cut-off value was 3.58 g/L, resulting in sensitivity, specificity, PPV, and NPV of 82.93%, 72.22%, 77.3%, and 78.1%, respectively (Table 7). From the above results, the value of PCT for diagnosing PJI was inferior to other markers. As shown in Table 7, PCT with the lowest Youden's index (0.02846) has the lowest PLR (1.17) but the highest NLR (0.97). Therefore, the same conclusion can be drawn from Youden's index, positive likelihood ratio (PLR), and negative likelihood ratio (NLR).

PJI is a catastrophic complication after total joint arthroplasty (20). Currently, there is an international consensus for the diagnosis of PJI, but no “gold standard” (21). The differentiation between PJI and aseptic loosening (AL) is challenging in orthopedic surgery because the treatment of PJI is entirely different to the treatment of aseptic loosening (22). ESR and CRP are initial markers recommended by the current guidelines due to the low false-negative and high sensitivity rates (23–25). CRP is a protein made by the liver. When there is acute inflammation, it responds to increased macrophages (26). Erythrocyte sedimentation rate (ESR) refers to the rate at which red blood cells sink under certain conditions. Although CRP and ESR have shown abilities for diagnosing PJI after primary replacement, the efficacy is limited (27, 28). Research by Paziuk T et al. showed that initial platelet (PLT) counts could be used for distinguishing between PJI and AL (29). Related studies reported that fibrinogen (FIB) with high sensitivity and specificity might become a novel biomarker for diagnosing PJI (19, 30).

PCT, which is produced by Thyroid C cells, consists of 116 amino acids. During infection, serum PCT levels rise with the bacterial endotoxin (31). Therefore, it is helpful to diagnose systemic infection (32). PCT is an undefined biomarker for diagnosing local infections such as PJI (33). The reports about PCT as a marker for the diagnosis of PJI are controversial. Sa-Ngasoongsong et al. showed that PCT was a very specific marker with low sensitivity for diagnosing PJI (34). In contrast, Glehr et al. demonstrated PCT is a sensitive, but not specific marker for diagnosis of PJI (35). Yoon et al. come to a conclusion that PCT is not a promising maker for diagnosing PJI (36). After reviewing reports for the diagnosis of PJI, we found Several studies have illustrated the role of PCT in diagnosing patients with PJI. These studies showed PCT is a sensitive and specific marker of bacterial infection (37–39). We want to investigate whether PCT is a superior maker. Therefore, we performed a sensitivity analysis to further validate the diagnostic performance of PCT. Finally, we found different results which showed the limited efficacy of PCT for diagnosing PJI before revision arthroplasties.

There was no significant difference for PCT between PJI group and the AL group (P = 0.747). The AUC for PCT was 0.504 (95% CI 0.373–0.635, p = 0.950). The serum PCT indicated a sensitivity of 19.51% and a specificity of 83.33% for diagnosing PJI. When the optimal cut-off value for PCT was set as 0.05 ng/ml, its PPV and NPV were 57.1% and 47.6%, respectively. The results in our study showed that PCT is a specific, but less sensitive biomarker for diagnosing PJI. The AUCs of other biomarkers increased significantly as they were combined with PCT. In conclusion, the combination of serum PCT with one of the other markers can improve their diagnostic accuracies.

Different reasons may result in the limited efficacy of PCT for diagnosing PJI. Firstly, PCT is not necessarily released into the blood if patients suffering from PJI do not show bacteremia (40). It is conceivable that the grade and virulence of the majority of PJI is too low to trigger PCT release. It was pointed out that the high rate of false negatives is associated with local low-virulence organisms (41). Secondly, in healthy adults, even after tooth brushing, transient bacteremia may lead to low-grade PCT release (42–44). Thirdly, since the penetration of PCT into the blood is different in each patient, the cut-off values set for the PCT may affect the study results. The cut-off value (0.05 ng/ml) may not be optimal in our cohort. In addition, we performed only the measurement of serum PCT, while we did not conduct the measurement of synovial fluid PCT.

Some limitations should be considered in our study. Firstly, the modified MSIS criteria used in this study may produce bias for assessing the diagnostic accuracy. Secondly, this was a retrospective cohort study, so its own inherent selection bias may affect its results. Thirdly, only 77 patients were recruited for the study. Thus, a multi-center study with a larger sample size is needed to be carried out for further analyses. This also diminished the sample size. Finally, our study has shown only serum PCT biomarkers, not including synovial fluid PCT biomarkers.

We detected multiple biomarkers for their diagnostic performance. In conclusion, this study demonstrates that the value of serum PCT has limited efficacy in differentiating PJI from aseptic loosening before revision arthroplasties. However, PCT combined with other biomarkers further increases diagnostic accuracy. Further multiple-center studies with large-size samples are needed to improve its diagnostic rate and validate our results.