Galactomannan and Polymerase Chain Reaction-Based Screening for Invasive Aspergillosis Among High-Risk Hematology Patients: A Diagnostic Meta-analysis

Background. Screening of high-risk patients for invasive aspergillosis (IA) has the potential to decrease the use of empiric antifungal agents. However, the performance of different screening methods has not been studied. Methods. We performed a meta-analysis of published studies to assess the diagnostic performance of galactomannan (GM) and polymerase chain reaction (PCR) as weekly screening tests in high-risk populations. The sensitivity and specificity of 6 approaches combining GM and PCR were estimated using the bivariate model. Results. Thirteen studies with 1670 patients met our inclusion criteria. Single positive test results had modest sensitivity and specificity for screening (respectively, 92% and 90% for GM; 84% and 76% for PCR). The screening approach with the highest sensitivity was the one that used at least 1 GM-or PCR-positive result to define a positive episode, achieving a sensitivity of 99%, significantly higher than any single test (P = .0018 compared with GM and P < .0001 compared with PCR). Meanwhile, when both GM and PCR were positive for the same patient, the specificity increased to 98%, which was not significantly different compared to the specificity of at least 2 positive GM (95%, P = .56 for the comparison) or PCR results (93%, P = .07 for the comparison). Conclusions. When screening high-risk patients for IA with GM and PCR tests, the absence of any positive test can obviate the need for antifungal agents with a negative predictive value of 100%, whereas the presence of at least 2 positive results is highly suggestive of an active infection with a positive predictive value of 88%.