Invasive pulmonary aspergillosis (IPA) is a major cause of mortality in patients with stem cell transplants and hematologic malignancies. Timely diagnosis of IPA improves survival but is difficult to make. We evaluated the effectiveness of bronchoalveolar lavage (BAL) galactomannan (GM) in diagnosing IPA in these populations by retrospectively reviewing records of 67 consecutive patients, in whom 89 BAL GM tests were performed. For patients with IPA, only the first BAL sample linked to the IPA episode was analyzed. Eighty samples were associated with proven, 12 with probable, and 32 with possible invasive fungal infections (IFI), and 37 were associated with no IFI. Among patients with IFIs, 4 had proven, 11 probable, and 32 possible IPA. Using BAL GM ≥ 0.5 (cutoff for serum GM) and ≥ 0.85 (optimal cutoff identified by receiver-operating characteristic curve), the sensitivity in diagnosing proven or probable IPA was 73% (11/15) and 67% (10/15), respectively, and specificity was 89% (33/37) and 95% (35/37). At these cutoffs, positive and negative predictive values were 73% (11/15) and 83% (10/12), and 89% (33/37) and 87% (35/40), respectively. BAL GM was more sensitive than cytology (0%, 0/14), BAL culture (27%, 4/15), transbronchial biopsy (40%, 2/5), or serum GM (67%, 10/15) for diagnosing IPA. BAL GM was ≥ 0.85 and ≥ 0.5 in 86% (6/7) and 100% (7/7) of patients with proven or probable IPA who received a mold-active agent for ≤ 3 days. BAL GM added sensitivity to serum GM and other means of diagnosing IPA, and was not impacted by short courses of mold-active agents.
2011 American Society for Blood and Marrow Transplantation. All rights reserved.