Julio C. Zuniga-Moya (@julioczuniga1.bsky.social)

Sensitivity and Specificity of Polymerase Chain Reaction in Blood and Bronchoalveolar Lavage Samples for Mucormycosis: A Bayesian Diagnostic Test Accuracy Meta-Analysis - Mycopathologia Background The utility of polymerase chain reaction (PCR) for diagnosing mucormycosis is uncertain. We conducted a diagnostic test accuracy meta-analysis to determine the performance of Mucorales PCR. Methods We systematically reviewed PubMed, Embase, and Cochrane Central databases to identify studies reporting Mucorales PCR (blood or bronchoalveolar lavage fluid [BALF] samples) in at least 10 suspected mucormycosis cases. We pooled sensitivity and specificity using Bayesian (weak and two informative priors) and frequentist methods. Subgroup analyses included middle vs. high-income countries, risk factors (hematological diseases vs. others), study design, assay type (commercial vs. in-house), and testing protocol (serial testing vs. single-time PCR). We assessed the effect of the number of blood tests performed per participant on sensitivity using multivariable meta-regression. Results We identified 1524 citations, of which 36 studies (41 datasets, n = 6627 participants; 16,613 samples) were included. The pooled sensitivity and specificity (with 95% credible interval) by Bayesian weak priors (primary analysis) were 51% (34–67%) and 97% (95–98%) for blood; 64% (50–76%) and 97% (95–98%) for BALF samples. We found wide variation in sensitivity, with a 14% increment for every additional blood sample tested per patient. Studies from middle-income countries reported a significantly lower pooled sensitivity (31 vs. 68%) in blood samples and lower specificity in BALF (72 vs. 98%) than high-income countries. No other subgroups showed significant differences. Conclusion Mucorales PCR was highly specific but poorly sensitive in blood, and moderately sensitive with high specificity in BALF. Before routine implementation, further well-designed studies from diverse settings and underrepresented regions are needed. Graphical Abstract https://link.springer.com/article/10.1007/s11046-025-00996-w