Does the Alpha Defensin ELISA Test Perform Better Than the Alpha Defensin Lateral Flow Test for PJI Diagnosis? A Systematic Review and Meta-analysis of Prospective Studies

Periprosthetic joint infection (PJI) following total joint arthroplasty is a serious complication that causes severe morbidity and adds a major financial burden to the healthcare system. Although there is plenty of research on the alpha-defensin (AD) test, a meta-analysis consisting of only prospective studies investigating AD's diagnostic efficacy has not been performed. Additionally, some important subgroups such as THA and TKA have not been separately analyzed, particularly regarding two commonly used versions of the AD test, the laboratory-based (ELISA) and lateral-flow (LF). (1) Does the AD ELISA test perform better in the detection of PJI than the AD LF test, in terms of pooled sensitivity and specificity, when including prospective studies only? (2) Are there differences in sensitivity or specificity when using AD ELISA and AD LF tests for PJI diagnosis of THA or TKA PJI separately? Following the Preferred Reporting Items for Systematic reviews and Meta-Analyses guidelines, we included prospective studies describing the use of either AD test in the workup of pain after total joint arthroplasty (primary or revision, but not after resection arthroplasty). Fifteen studies (AD ELISA: 4; AD LF: 11) were included, with 1592 procedures. Subgroup data on THA and TKA could be retrieved for 1163 procedures (ELISA THA: 123; LF THA: 257; ELISA TKA: 228; LF TKA: 555). Studies not describing THA or TKA, those not using Musculoskeletal Infection Society (MSIS) criteria as the standard for determining the presence or absence of PJI, those not clearly reporting data for the AD test for the total cohort, and those describing data published in another study were excluded. Studies were not excluded based on follow-up duration; the MSIS criteria could be used within a few weeks, when test results were available. Quality was assessed using the Quality Assessment of Diagnostic Accuracy Studies-2 criteria. Study quality was generally good. The most frequent sources of bias were related to patient selection (such as unclear inclusion and exclusion criteria) and flow and timing (uncertainty in place and time of aspiration, for example). Heterogeneity was moderate to high; a bivariate random-effects model therefore was used. To answer both research questions, sensitivity and specificity were calculated for AD ELISA and LF test groups and THA and TKA subgroups, and were compared using z-test statistics and meta-regression analysis. No differences were found between the AD ELIS