Genes that encode diverse types of immunostimulatory molecules can enhance antitumor immunity when introduced into tumor cells, but this strategy rarely leads to rejection of preexisting tumors. Lymphotactin, a recently discovered chemokine, induced T-cell migration in vitro in the absence of any chemotactic effects on monocytes for granulocytes. Thus, combined delivery of lymphotactin and a cytokine such as IL2, to the vicinity of an antigen-expressing tumor cell, could be expected to accelerate and potentiate the antitumor response over that elicited by either agent alone. This prediction has been substantiated in the applicant's laboratory in a murine model (Dilloo et al. Nature Medicine, in press) leading to the current proposal to investigate lymphotactin/IL2 immunomodulation in children with relapsed or refractory neuroblastoma (stage C or D). Preliminary results of a clinical gene transfer protocol at this institution indicate that neuroblastoma cells transduced with an IL2 adenovirus are safe to deliver to children, will elicit both local and systemic antitumor responses, and may induce objective clinical responses. The present application attempts to build on this experience by testing the safety and potential immunogenicity of neuroblastoma cells transduced with lymphotactin, using the same adenoviral vector that was so effective in studies with IL2 (Aim 1). Once the optimal therapeutic concentration of IL2-secreting cells is determined, these neuroblasts will be injected (at a fixed dose) with escalating doses of lymphotactin-secreting cells to patients with drug-resistant disease (aim 2). Comparison of local and systemic antitumor immune responses with those elicited by either agent alone will reveal whether the lymphotactin/IL2 combination has significant immunostimulatory advantages in vivo. The results obtained with neuroblastoma cells should be generally applicable with other human cancers.