Toxicity of human recombinant interleukin-2 in the mouse is mediated by interleukin-activated lymphocytes. Separation of efficacy and toxicity by selective lymphocyte subset depletion.

Human recombinant interleukin-2 (rIL-2) was administered to normal and tumor-bearing BDF mice for 1 to 3 weeks, and the hematologic, clinical chemistry, gross and histopathologic findings were evaluated. Vascular leak syndrome (pulmonary edema, pleural effusion, ascites), hepatocyte necrosis, elevated hepatic serum transaminases, hypoalbuminemia, tissue and peripheral eosinophilia, thrombocytopenia, and prerenal azotemia were the detrimental effects of rIL-2 treatment. Vascular leak syndrome and hepatocyte necrosis were causally associated with vascular-oriented lymphocytic infiltration of pulmonary and hepatic parenchyma. Pleural effusions contained up to 99,000 cells/mm3, most of which were large granular lymphocytes. Antiserum to the glycolipid asialo GM1 (ganglio-n-tetrosylceramide), given simultaneously with rIL-2, prevented overt toxicity of rIL-2 (mortality, vascular leak syndrome, and hepatic damage) and substantially reduced infiltration of pulmonary and hepatic vasculature by asialo GM1+ lymphocytes. Asialo GM1 antiserum did not inhibit lymphoid hyperplasia, tissue infiltration by Lyt 2+ lymphocytes, tissue and peripheral eosinophilia, or thrombocytopenia in rIL-2 treated mice. Additionally, asialo GM1 antisera prevented toxicity, but not anti-tumor efficacy, of high dose rIL-2 therapy in BDF mice bearing the colon 38 adenocarcinoma. These results suggest that, in BDF mice and with this tumor model, vascular leak syndrome and hepatocyte necrosis are mediated by an endogenous subset of rIL-2-stimulated lymphocytes which are asialo GM positive, that mechanisms of toxicity and efficacy associated with high dose rIL-2 therapy are not necessarily the same, and that these mechanisms can be therapeutically separated.