A preclinical pilot study was done to evaluate the effects of a continuous regional hepatic arterial infusion of human recombinant interleukin-2 (IL-2) in dogs with an infusion pump. Preliminary studies demonstrated the ability to culture canine lymphokine-activated killer (LAK) cells in vitro and a canine LAK cell 15Cr assay was developed with a canine tumor cell line (CTAC) with appropriate controls. An in vitro study of the stability of IL-2 in the pump was done with a bioassay and enzyme-linked immunosorbent assay for IL-2 that demonstrated the stability of IL-2 during a 14-day period at 37 degrees C. Infusions of 300, 600, and 1200 units/kg/hr IL-2 were tested in vivo in dogs. LAK cell and natural killer cell activity, blood counts, and hepatic and renal function were monitored for 1 month. No significant natural killer or LAK response or toxicity was found at the 300 unit/kg/hr level. Infusion of 600 units/kg/hr was associated with a significant increase of the cytotoxic activity of peripheral blood lymphocytes after 3 weeks of treatment. At the 1200 unit/kg/hr level, increased activity occurred at 1 week and thereafter. The only significant toxicity was a 15% increase in body weight occurring during the infusion of 1200 units/kg/hr. Results of renal and hepatic function studies remained normal except for a slight elevation of transaminase levels after 4 weeks of 1200 units/kg/hr. A significant rise in eosinophil count was noted at each dosage level. Results of autopsies were unremarkable. These data demonstrate that continuous hepatic arterial regional infusion with relatively low doses of IL-2 is able to stimulate a sustained in vivo peripheral blood LAK cell effect in dogs with the absence of major side effects. These findings suggest that these methods may have both research application in large animals and clinical application in patients with tumors that are responsive to LAK cell lysis.