Diagnostic accuracy of PCR alone compared to galactomannan in bronchoalveolar lavage fluid for diagnosis of invasive pulmonary aspergillosis: a systematic review.

PCR in bronchoalveolar lavage (BAL) fluid has not been accepted as a diagnostic criterion for invasive pulmonary aspergillosis (IPA). We conducted a systematic review assessing the diagnostic accuracy of PCR in BAL fluid with a direct comparison versus galactomannan (GM) in BAL fluid. We included prospective and retrospective cohort and case-control studies. Studies were included if they used the EORTC/MSG consensus definition criteria of IPA and assessed ≥80% of patients at risk for IPA. Two reviewers abstracted data independently. Risk of bias was assessed using QUADAS-2. Summary sensitivity and specificity values were estimated using a bivariate model and reported with a 95% confidence interval (CI). Nineteen studies published between 1993 and 2012 were included. The summary sensitivity and specificity values (CIs) for diagnosis of proven or probable IPA were 90.2% (77.2 to 96.1%) and 96.4% (93.3 to 98.1%), respectively. In nine cohort studies strictly adherent to the 2002 or 2008 EORTC/MSG criteria for reference standard definitions, the summary sensitivity and specificity values (CIs) were 77.2% (62 to 87.6%) and 93.5% (90.6 to 95.6%), respectively. Antifungal treatment before bronchoscopy significantly reduced sensitivity. The diagnostic performance of PCR was similar to that of GM in BAL fluid using an optical density index cutoff of 0.5. If either PCR or GM in BAL fluid defined a positive result, the pooled sensitivity was higher than that of GM alone, with similar specificity. We conclude that the diagnostic performance of PCR in BAL fluid is good and comparable to that of GM in BAL fluid. Performing both tests results in optimal sensitivity with no loss of specificity. Results are dependent on the reference standard definitions.