Clinical utility of bronchoalveolar lavage in immunocompromised hosts

Bronchoalveolar lavage (BAL) has been used extensively for assessment of immunocompromised hosts with pulmonary infiltrates. Reported estimates of the diagnostic utility of BAL have varied because of differences in patient populations, diagnostic criteria, and study methods. Herein we report on the use of BAL to determine at least one of the final diagnoses in 150 immunocompromised patients. Although the frequency with which BAL provided at least one of the final diagnoses (overall diagnostic yield) was seemingly low (39%), the yield increased substantially when only patients with pathologically proven diagnoses were considered. The sensitivity of BAL was 82%, and the specificity was 53%. The use of rigid diagnostic criteria enabled us to distinguish pathogens from colonizers. Pneumocystis was considered a pathogen whenever it was identified. It was the most common infectious pathogen identified (50%) despite the fact that our study population had relatively few patients (only 4%) with acquired immunodeficiency syndrome (AIDS). Organisms such as cytomegalovirus, Aspergillus, and Candida were frequently identified in BAL specimens but were eventually proved to be pathogens in only 24%, 25%, and 0% of cases, respectively. BAL detected pulmonary malignant lesions on the basis of positive cytologic results in four of six patients eventually found to have primary or metastatic lung cancer. Our results should enhance the understanding of the strengths and weaknesses of BAL and assist in the interpretation of associated microbiologic findings.