Lung biopsy in immunocompromised patients: one institution's experience and an approach to management of pulmonary disease in the compromised host.

The authors report on their institution's experience with 53 lung biopsies, including 26 open, 22 transbronchial, and five trephine air drill biopsies, performed in immunocompromised patients with roentgenographic pulmonary infiltrates. Open biopsy was far more likely to provide a specific etiologic diagnosis (81%, P less than 0.001) than transbronchial biopsy (32%), or trephine biopsy (20%). Infection (17 biopsies), neoplastic disease, (7) or drug-related pneumonitis (2) were identified most frequently. Patients with myeloproliferative disease, granulocytopenia, or those who had not received prior immunosuppressive therapy were most likely to have a nondiagnostic biopsy (P less than 0.05 for each factor). The overall complication rate of biopsy procedures was 15% and was comparable with all three methods. Survival in this series was not significantly lower if a specific etiologic diagnosis could not be established, but correlated with the respiratory rate (less than 20 per minute), pO2 (greater than 60 torr), and the roentgenographic pattern (other than bilateral diffuse disease) at the time of biopsy (P less than 0.05 for each factor). There were 18 cases (34%) in which a clearcut etiologic diagnosis would not be established at the time of biopsy; based on serologic tests performed ex post facto, 2 of 12 of these cases (17%) were Legionnaire's disease. Lung biopsies were helpful in the management of the majority of the cases, although nondiagnostic biopsies continue to be a problem. The authors propose an approach to the management of compromised patients with pulmonary infiltrates.