Is Plasma D-Dimer Really as Reliable as CRP for Diagnosing PJI?: Commentary on an article by Saad Tarabichi, MD, et al.: "Plasma D-Dimer Is Noninferior to Serum C-Reactive Protein in the Diagnosis of Periprosthetic Joint Infection".

Abstract

Commentary The most important issues with respect to the diagnosis of periprosthetic joint infections (PJIs) are to establish a clear definition of the diagnostic criteria and to determine the threshold of each test based on strong evidence. At the first International Consensus Meeting (ICM) in 2013, the diagnostic criteria for PJI were established on the basis of serological markers such as C-reactive protein (CRP), microbiological cultures, several joint fluid tests, and histological findings. The bases of these criteria remain the same and are referred to by the majority of PJI-related studies. At the 2018 ICM, a major change was made in that D-dimer was included as one of the minor criteria. At that point, there was insufficient evidence for PJI diagnosis based on D-dimer levels, particularly regarding the optimal threshold. Most studies about D-dimer and PJI have been published after 2018. Several recent systematic reviews and meta-analyses have shown an acceptable pooled sensitivity of approximately 0.73 to 0.81 and specificity of 0.74 to 0.78 for D-dimer1–3, using thresholds between 756 and 1,250 ng/mL1. The current study by Tarabichi et al., which includes the greatest number of subjects tested in a prospective setting, compares the diagnostic accuracy of D-dimer with that of the CRP level. As their gold-standard reference, the authors modified the 2018 ICM criteria by excluding D-dimer (as well as CRP and the erythrocyte sedimentation rate [ESR]) as a serological marker. An important finding was that the accuracy of serum D-dimer was comparable with that of CRP when using the optimal cutoff value of 664 ng/mL; notably, this is lower than most previously reported thresholds. Indeed, receiver operating characteristic (ROC) analysis revealed that the area under the curve (AUC) for D-dimer (0.860) was almost the same as that for CRP (0.862). Furthermore, the best AUC for D-dimer was achieved after excluding patients with comorbidities such as systemic inflammatory diseases, active malignancy, or a history of venous thromboembolism. A strength of this study based on a large prospective cohort is the generalizability of a D-dimer-based diagnosis to a PJI population with various backgrounds. The heterogeneity of PJI status meant that various bacteria were identified in their cohort, as were culture-negative cases. This is quite important to ensure diagnostic generalizability. It is also notable that they performed subgroup analyses of patients with various conditions (reported in their Appendix). Two of the subgroup analyses, involving the assignment of patients with an inconclusive ICM score to the infected group and the assessment of data obtained when a diagnosis of PJI was caused by indolent organisms, are of special interest. Such detailed subgroup analyses were possible because of the large number of subjects and the lack of missing data due to the prospective study design. A major limitation of this study was the gold-standard definition of PJI, a problem commonly associated with diagnostic studies. Because the 2018 ICM criteria already include D-dimer, CRP, and ESR, the authors had to exclude these data when calculating ICM scores. This is inevitable, albeit somewhat contentious, because the modified 2018 ICM criteria that exclude D-dimer, CRP, and ESR are not validated, as the authors also noted. Nevertheless, I believe that their findings provide supporting evidence for the clinical application of D-dimer for diagnosing PJI as reliable as CRP. Particularly when considering the optimum threshold for plasma D-dimer in future updates of the ICM criteria, current information should be cited. Further studies involving multiple centers are needed to strengthen the evidence for D-dimer as a diagnostic criterion for PJI.