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

Summary 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.