BACKGROUND Despite vigorous efforts at curbing tobacco consumption and aggressive combined-modality treatment programs, both the incidence of and the mortality from lung cancer have remained virtually unchanged in the last 10 years. More effective innovative therapies are clearly needed. The direct transfer into tumor cells of tumor suppressor genes or toxic gene products that specifically promote tumor cell death and spare nonmalignant cells is a potentially novel anticancer treatment approach that should be investigated. OBJECTIVE On the basis of compelling preclinical data, we initiated a phase I study involving six patients with inoperable lung cancer and an endobronchial lesion accessible by bronchoscopy. Our purpose was to evaluate the feasibility, tolerance, and clinical, biologic, and immunologic effects of the intratumoral administration of a recombinant, replication-deficient adenovirus (rAd.RSV beta-gal), using the Rous sarcoma virus promoter to drive transcription of the Escherichia coli lacZ marker gene that encodes for the bacterial enzyme beta-galactosidase (beta-gal). METHODS From June 1994 through April 1995, six patients (five males and one female) were enrolled in the trial. A single dose of recombinant virus suspension containing 10(7) or 10(8) plaque-forming units (PFU) was injected intratumorally into two successive cohorts of three patients. Eligible patients received concomitant chemotherapy. Patients were kept under isolation conditions from 3 days before the injection was given until virus excretion was undetectable. Biopsy specimens of the tumor and surrounding mucosa were collected on the 8th day and at 1, 2, and 3 months after injection. They were analyzed by cell culture, polymerase chain reaction (PCR), and beta-gal expression for the presence of recombinant adenovirus. So that the risk of virus recombination or complementation could be minimized, wildtype adenovirus carriers among the hospital staff (identified by PCR) were excluded from contact with patients who were potentially excreting recombinant virus. RESULTS beta-gal was expressed in tumor biopsy specimens of three patients (one who received the 10(7) PFU dose level and two who received 10(8)). Bronchoalveolar lavage specimens collected immediately after injection were positive for recombinant adenovirus when analyzed in culture and by PCR. All biologic fluids were negative for recombinant virus as judged by PCR after day 12, with the exception of bronchoalveolar lavage specimens (positive PCR up to 90 days in two of three patients treated with 10(8) PFU). The blood samples obtained from the three patients treated with 10(8) PFU showed positive PCR results immediately after virus injection. Patients were kept in isolation for a median of 17 days. The most common toxic effects were moderate bleeding (occurring in two patients) during bronchoscopy and fever (seen in four patients). Endoscopic and clinically objective antitumor responses were seen in four patients, including one patient who showed a complete response by pathologic evaluation. The median survival for the patients was 12.5 months (range, 3-16+ months). Throughout the study, hospital staff remained negative for recombinant adenovirus infection. CONCLUSIONS This ongoing phase I study has demonstrated that a recombinant adenovirus-mediated marker gene, such as rAd.RSV beta-gal, can be safely introduced into humans and that the gene product is expressed by lung tumor cells of the host.