Abstract
INTRODUCTION: The diagnosis of invasive pulmonary aspergillosis, according to the criteria as defined by the European Organisation for the Research and Treatment of Cancer/Mycoses Study Group (EORTC/MSG), is difficult to establish in critically ill patients. The aim of this study is to address the clinical significance of isolation of Aspergillus spp. from lower respiratory tract samples in critically ill patients on the basis of medical and radiological files using an adapted diagnostic algorithm to discriminate proven and probable invasive pulmonary aspergillosis from Aspergillus colonisation. METHODS: Using a historical cohort (January 1997 to December 2003), all critically ill patients with respiratory tract samples positive for Aspergillus were studied. In comparison to the EORTC/MSG criteria, a different appreciation was given to radiological features and microbiological data, including semiquantitative cultures and direct microscopic examination of broncho-alveolar lavage samples. RESULTS: Over a 7 year period, 172 patients were identified with a positive culture. Of these, 83 patients were classified as invasive aspergillosis. In 50 of these patients (60%), no high risk predisposing conditions (neutropenia, hematologic cancer and stem cell or bone marrow transplantation) were found. Typical radiological imaging (halo and air-crescent sign) occurred in only 5% of patients. In 26 patients, histological examination either by ante-mortem lung biopsy (n = 10) or necropsy (n = 16) was performed, allowing a rough estimation of the predictive value of the diagnostic algorithm. In all patients with histology, all cases of clinical probable pulmonary aspergillosis were confirmed (n = 17). Conversely, all cases classified as colonisation had negative histology (n = 9). CONCLUSION: A respiratory tract sample positive for Aspergillus spp. in the critically ill should always prompt further diagnostic assessment, even in the absence of the typical hematological and immunological host risk factors. In a minority of patients, the value of the clinical diagnostic algorithm was confirmed by histological findings, supporting its predictive value. The proposed diagnostic algorithm needs prospective validation.