A specific histologic diagnosis was obtained in 62 (71%) of the patients, 33 of whom had infections. Pneumocystis carinii was the most common microbial pathogen (16 patients), but no cases have been observed since 1980 when the routine use of prophylactic trimethoprim/sulfa began. The other specific diagnoses included malignancy or drug-induced lung disease. Specific therapy was available for 52 patients, and in 33 cases, a change in therapy was necessary to treat according to the lung biopsy diagnosis. Forty-one patients received an adequate course of therapy and 27 (66%) of these improved clinically, including 16 of 26 patients with infections, 11 of 14 with malignancies, and 1 of 2 with a vasculitis. Among the subgroup of 33 patients for whom a new, specific therapeutic option was available as a result of the biopsy diagnosis 21 (64%) responded to the treatment. Eleven significant operative complications were encountered, but no deaths were attributable to the biopsy. An open-lung biopsy in immunocompromised patients is a relatively safe, accurate diagnostic procedure which frequently facilitates appropriate therapy and clinical improvement. Cancer 55:453-459. 1985. s MANY AS 60% of immunocompromised patients A Twill develop pulmonary infiltrates,' and these cases present a diagnostic and therapeutic challenge to clinicians. When such a patient fails to respond to empiric therapy, a variety of invasive, diagnostic procedures are often considered to provide tissue for examination. Those procedures which have been recommended most freque:ntly include the transthoracic needle aspiration or b i o p ~y , ~-~ bronchoscopy with transbronchial b i o p ~y , ~, ~ and open-lung biopsy.
'O-l6
There is disagreement among the published series concerning the value of an open-lung biopsy (Table 1). In ten previous reports of lung biopsy findings in im-