• Jorien D'Haese, Koen Theunissen, Edith Vermeulen, Hélène Schoemans, Greet De Vlieger, Liesbet Lammertijn, Philippe Meersseman, Wouter Meersseman, Katrien Lagrou, and Johan Maertens.
• Department of Hematology, Acute Leukemia and Stem Cell Transplantation Unit, Universitaire Ziekenhuizen Leuven, Campus Gasthuisberg, Catholic University of Leuven, Leuven, Belgium.
• J. Clin. Microbiol. 2012 Apr 1; 50 (4): 1258-63.

AbstractInvasive pulmonary aspergillosis (IPA) is frequent and often fatal in immunosuppressed patients. Timely diagnosis of IPA improves survival but is difficult to make. We examined the analytical and clinical validity of galactomannan (GM) testing of bronchoalveolar lavage (BAL) fluid in diagnosing IPA in a mixed population by retrospectively reviewing records of 251 consecutive at-risk patients for whom BAL fluid GM testing was ordered. The performance of the enzyme immunoassay was evaluated by using a range of index cutoffs to define positivity. Three samples were associated with proven IPA, 56 were associated with probable IPA, 63 were associated with possible invasive fungal disease (IFD), and 129 were associated with no IFD. Using a BAL fluid GM index of ≥0.8 (optimal optical density [OD] index cutoff identified by a receiver operating characteristic curve), the sensitivity in diagnosing proven and probable IPA was 86.4%, and the specificity was 90.7%. At this cutoff, positive and negative predictive values were 81% and 93.6%, respectively. However, an OD index value of ≥3.0 corresponded to a 100% specificity, thus ruling the disease in, irrespective of the pretest probability. Conversely, an OD index cutoff of <0.5 corresponded to a high sensitivity, virtually always ruling the disease out. For all values in between, the posttest probability of IPA depends largely on the prevalence of disease in the at-risk population and the likelihood ratio of the OD index value. Detection of GM in BAL fluid samples of patients at risk of IPA has an excellent diagnostic accuracy provided results are interpreted in parallel with clinico-radiological findings and pretest probabilities.

20 keywords...

hide…