False positive bronchoalveolar lavage galactomannan: Effect of host and cut-off value

Objectives Bronchoalveolar lavage galactomannan (BAL-GM) is a mycological criterion for diagnosis of probable invasive aspergillosis (IA) per European Organization for Research and Treatment of Cancer/Mycoses Study Group (EORT-MSG) consensus criteria, but its real-world positive predictive value (PPV) has not been well-studied. Our aim was to estimate the PPV of BAL-GM in a contemporary cohort of patients with positive BAL-GM. Methods We identified consecutive patients with >= 1 positive BAL-GM value (index >= 0.5) at Brigham and Women's Hospital from 11/2009 to 3/2016. We classified patients as having no, possible, probable, or proven IA, excluding BAL-GM as mycological criterion. Results We studied 134 patients: 54% had hematologic malignancy (HM), and 10% were solid organ transplant (SOT) recipients. A total of 42% of positive (>= 0.5) BAL-GM results were falsely positive (PPV 58%). The number of probable IA cases was increased by 23% using positive BAL-GM as mycologic criterion alone. PPV was higher in patients with HM or SOT (P < 0.001) and with use of higher thresholds for positivity (BAL-GM >= 1 vs 1-0.8 vs 0.8-0.5: P = 0.002). Conclusions 42% of positive BAL-GM values were falsely positive. We propose a critical reassessment of BAL-GM cutoff values in different patient populations. Accurate noninvasive tests for diagnosis of IA are urgently needed.