A graft-versus-leukemia (GVL) effect has been considered a major factor responsible for cures in patients with hematologic malignancies undergoing allogeneic bone marrow transplantation; however, associated graft-versus-host disease (GVHD) results in significant morbidity and mortality. T-cell depletion reduces the incidence and severity of GVHD but eliminates, at least partially, the GVL effect. Reinfusion of donor T lymphocytes at relapse posttransplantation can induce a potent antitumor response, but GVHD still occurs in the majority of patients. Prior transduction of T lymphocytes with the suicide gene, the viral thymidine kinase (TK), permits specific cell kill on administration of ganciclovir (GCV). Therefore, infusion of TK-transduced T lymphocytes may induce GVL effect and allow for their subsequent selective elimination in case GVHD develops. To evaluate the efficacy and feasibility of this promising approach, anti-CD3-stimulated primary human lymphocytes cultured in interleukin-2 were TK-transduced by a retroviral vector carrying both TK and neomycin-resistance genes. After selection in G418, more than 90% of the cells contained the TK gene as shown by a semiquantitative polymerase chain reaction. In addition, 1 to 5 days of GCV exposure, at clinically achievable concentrations of 20 to 50 micromol/L, induced > or = 90% killing of G418-selected cells without affecting nontransduced cells. Correlation of the extent of T-cell kill and the proportion of TK-gene-transduced cells is consistent with the absence of a bystander effect. Transduced cells were CD3+ and either CD8+ or CD4+ and retained functional properties of untransduced cells. In vivo administration of GCV prevented tumor development after subcutaneous injection of TK-transduced murine myeloma cells (MOPC-11), whereas such an effect was not observed on injection of untransduced cells into the opposite flank. Our studies provide critical information that (1) adequate numbers of TK-transduced lymphocytes can be selected efficiently with > or = 90% purity, (2) selected cells remain functional, (3) 24 hours of exposure to GCV at clinically achievable concentration effects > or = 90% killing of selected cells, and (4) GCV is effective in vivo in killing TK-transduced cells. Based on these data, a clinical study has been initiated in patients with multiple myeloma with persistent or relapsing disease after T-cell-depleted allogeneic transplants.