The potential efficacy and clinical feasibility of gene therapy for prostate cancer were tested. Efficacy was tested using the Dunning rat prostate carcinoma model. Rats with anaplastic, hormone refractory prostate cancer treated with irradiated prostate cancer cells genetically engineered to secrete human granulocyte-macrophage colony-stimulating factor (GM-CSF) showed longer disease-free survival compared to either untreated control rats or rats receiving prostate cancer cell vaccine mixed with soluble human GM-CSF. A gene modified prostate cancer cell vaccine thus provided effective therapy for anaplastic, hormone refractory prostate cancer in this animal model. An evaluation of the clinical feasibility of gene therapy for human prostate cancer based on these findings was then undertaken. Prostate cancer cells from patients with stage T2 prostate cancer undergoing radical prostatectomy were first transduced with MFG-lacZ, a retroviral vector carrying the beta-galactosidase reporter gene. Efficient gene transfer was achieved in each of 16 consecutive cases (median transduction efficiency 35%, range 12 to 65%). Cotransduction with a drug-selectable gene was not required to achieve high yield of genetically modified cells. Histopathology confirmed malignant origin of these cells and immunofluorescence analysis of cytokeratin 18 expression confirmed prostatic luminal-epithelial phenotype in each case tested. Cell yields (2.5 x 10(8) cells per gram of prostate cancer) were sufficient for potential entry into clinical trials. Autologous human prostate cancer vaccine cells were then transduced with MFG-GM-CSF, and significant human GM-CSF secretion was achieved in each of 10 consecutive cases. Sequential transductions increased GM-CSF secretion in each of 3 cases tested, demonstrating that increased gene dose can be used to escalate desired gene expression in individual patients. These studies show a preclinical basis for proceeding with clinical trials of gene therapy for human prostate cancer.