Galactomannan antigen detection using bronchial wash and bronchoalveolar lavage in patients with hematologic malignancies

Mahnaz Taremi, Michael E. Kleinberg, Elizabeth W. Wang, Bruce L. Gilliam, Patrick A. Ryscavage

The diagnosis of invasive pulmonary aspergillosis is challenging. It is unclear whether galactomannan (GM) results from bronchial wash (BW) and bronchoalveolar lavage (BAL) samples differ in a clinically meaningful way. Ninety-six paired (BAL and BW) samples from 85 patients were included. The average age was 53 years, 61 % of the patients were male, and 74.1 % had an underlying diagnosis of AML/MDS (ALL 7.1 %, other hematologic malignancy 18.8 %). 57 (67.1 %) patients were neutropenic, and 56 (65.9 %) patients were receiving mold-active drugs at least 48 h prior to bronchoscopy. The overall agreement between GM detection from BW and BAL was 63.5 % (K = 0.152; 95 % CI 0.008-0.311) and 73 % (K = 0.149; 95 % CI 0.048-0.348) at cut off ≥0.5 and ≥1.0, respectively. Among 43 positive samples, using a GM cut-off of 0.5, 39 (90.5 %) were positive in BW samples whereas 12 (29.3 %) were positive in BAL samples. The median level of GM in BW (0.28) samples was significantly higher than in BAL (0.20) samples among 53 samples with negative results (P = 0.001). There was no statistically significant difference in the median GM values between the BW and BAL samples with positive results (P = 0.08). There was no significant difference in GM detection between samples with positive and negative results with regard to antifungal, beta lactam antibacterial treatment or neutropenia (60.5 vs 56.6 %; 53.9 vs 46 %; 65.1 vs 54.7 %, respectively). This retrospective study examining two collection techniques suggests that BW may have higher diagnostic yield compared to bronchoalveolar lavage for GM detection.