This study was carried to compare the accuracy of plasma D-dimer, plasma fibrinogen, ALB, GLB, and AGR with traditional inflammatory biomarkers (WBC, CRP, and ESR) in diagnosing tibial infected nonunion. Our data demonstrated that the blood-based biomarkers plasma fibrinogen and GLB were associated with infected tibial nonunion, and could be used as new biomarkers to diagnose infected tibial nonunion. Furthermore, the integrated analysis of four promising biomarkers (ESR, plasma fibrinogen, GLB, and AGR) provided more accurate and more specific evaluation of tibial infected nonunion than the four biomarkers individually.
ALB and GLB are two easily accessible and reliable biomarkers evaluated in in liver function tests that are routinely performed before surgery, and are able to be performed rapidly and economically. ALB is a negative acute phase protein, and is considered to be a biomarker of inflammation and nutritional status [26; 27; 28]. A meta-analysis conducted by Yuwen et al. reported that ALB levels < 3.5 g/dL was associated with an almost 2.5 fold increased risk of surgical site infection (SSI) in orthopaedics. However, Wang et al. reported that ALB was not useful for diagnosis of patients suspected of having PJI. Similarly, Ye et al. found that the levels of ALB were not associated with PJI. In our study, the diagnostic accuracy of ALB in tibial infected nonunion was poor, with an AUC of 0.671, sensitivity of 88.14%, specificity of 9.38%, PPV of 37.41%, and NPV of 56.25%. GLB is another major serum protein component that consists of antibodies and inflammatory cytokines, including complements, interleukin-6, and immunoglobulins [29; 30]. There is an inverse relationship between GLB and ALB in response to inflammation and infection, and the AGR could indicate an inflammatory state more accurately. GLB and AGR have been validated to have potential roles in the pathogenesis of inflammatory and infectious diseases in various studies [31; 32]. Moreover, recent studies have demonstrated that GLB and AGR are associated with PJI and may serve as potential adjuvant biomarkers in the diagnosis of PJI [19; 20]. In accordance with these results, we also observed that high globulin levels and low AGR were independently associated with the risk of tibial infected nonunion. ROC curve analysis revealed that globulin and AGR showed acceptable predictive value for the diagnosis of tibial infected nonunion. The diagnostic value of GLB (AUC = 0.805) was slightly lower than that of AGR (0.815), which were higher than 0.8. Both biomarkers demonstrated fair sensitivity (GLB 55.93%, AGR 57.63%), but high specificity (GLB 96.88%, AGR 92.71%), reducing the misdiagnosis rate of tibial infected nonunion. Moreover, the PPV of GLB (91.67%) was the highest; suggesting that GLB may be superior as a single biomarker in predicting the diagnosis of tibial infected nonunion.
Recently, it has been demonstrated that systemic and local infections result in fibrinolytic activity [33; 34]. Moreover, coagulation-related indicators, including D-dimer and fibrinogen, are useful diagnostic markers for PJI [35; 36; 37; 38; 39]. Serum D-dimer and plasma fibrinogen have performed well in the diagnosis of infected nonunion [21; 22]. In our study, plasma fibrinogen (AUC = 0.819) demonstrated good diagnostic performance in infected nonunion, similar to the results of another recent study . These data suggest that plasma fibrinogen may be a novel biomarker to diagnose tibial infected nonunion. However, the diagnostic value of plasma D-dimer was limited in our study, with a fair AUC of 0.776. Conversely, the D-dimer, assayed in serum rather than plasma, has proven to be useful for preoperative prediction of infected nonunion after ORIF . Some studies demonstrated that serum D-dimer is superior to plasma D-dimer in the diagnosis of PJI [40; 41]. Consequently, high-quality prospective studies that address these research gaps are needed to validate the use of D-dimer as a biomarker for tibial infected nonunion.
WBC, CRP, and ESR are the most commonly used biomarkers of tibial infected nonunion. Unfortunately, they are usually affected by other factors such as physiological stress, treatment, and other diseases [5; 42]. Peripheral WBC is frequently normal in low-grade infections and affords little diagnostic help. The diagnostic value of WBC was limited in our study, with a fair AUC of 0.599. The standard WBC biomarker may not provide enough information to clearly distinguish between infected nonunion and aseptic nonunion. On the contrary, ESR had good performance in diagnosis of suspicious infection with higher sensitivity, which is verified by past studies [43; 44].
In subgroup analysis, we found that the diagnostic accuracy of these novel biomarkers was good in patients with tibial infected nonunion who also had other comorbidities. Considering the small number of cases, a larger number of cases should be enrolled into evaluate the diagnostic accuracies of these biomarkers and reduce potential bias. For example, recent use of antibiotics might influence inflammatory biomarkers. Our subgroup analysis showed that there were significant differences between patients with infected nonunion who recently used antibiotics vs. those who had not in a number of biomarkers, including WBC, CRP, and ESR levels (P < 0.05); other biomarkers were not significantly different, including plasma D-dimer, plasma fibrinogen, ALB, GLB, and AGR (P > 0.05). Therefore, our data suggest that plasma fibrinogen, GLB, and AGR may be better biomarkers to diagnose patients with suspected infected nonunion who have recently used antibiotics.
There are several limitations to our study. Firstly, this study was retrospective, with inherent biases. Secondly, the sample size of this study is fairly small. Finally, the different comorbidities and application of antibiotics among the patients may have influenced the observed results of the diagnostic accuracy of the biomarkers in patients with infected nonunion. Therefore, multicenter, prospective, comparative studies with larger samples are required to more thoroughly determine the accuracy of these biomarkers for predicting tibial infected nonunion.
It was a secondary analysis that recursive partitioning analysis of the inflammatory marker predictive values for infected nonunion, and is limited by the relatively small sample size of the cohort included. In order to avoid external validity, we will calibrate and externally validate this model with a future large observational cohort of patients with suspected nonunion to inform clinical utility.