We conducted a single-center retrospective cohort study to determine the performance characteristics of the galactomannan (GM) assay in bronchoalveolar lavage (BAL) in patients with hematologic malignancies. Patients were classified as proven, probable, possible, or no invasive pulmonary aspergillosis (IPA), according to international guidelines. A total of 173 BAL samples from 145 patients were included. There were 5 proven, 7 probable, and 35 possible cases of IPA. Using a GM index cutoff of ≥ 0.5, the sensitivity, specificity, and positive and negative predictive values (PPV and NPV, respectively) of the BAL GM assay were 100%, 78%, 26%, and 100%, respectively. Using a GM index cutoff of ≥ 2.0, the sensitivity and NPV remained 100%, but specificity and PPV increased to 93% and 50%, respectively. The BAL GM assay is a highly sensitive screening test for IPA in patients with hematologic malignancies. Increasing the cutoff value to 2.0 would improve the performance of this assay.
Copyright © 2010 Elsevier Inc. All rights reserved.