Human recombinant interleukin-2 provokes acute pulmonary vascular endothelial injury in rabbits.

Abstract

Human recombinant interleukin-2 (rIL-2), with or without lymphokine-activated killer cells, has been shown to mediate tumor regressions in animals and in humans. A well-recognized adverse effect of rIL-2 is the development of a generalized vascular leak syndrome that involves the pulmonary microvasculature. We present a rabbit model for the study of rIL-2 pulmonary toxicity that closely reflects the human situation. Rabbits were treated with rIL-2 (Cetus) by two dose/schedule schemata. Separate control groups received rIL-2 excipient (Cetus) or 5% dextrose in water (D5W). In a short-term model, animals were treated with 10(6) Cetus units of rIL-2 in five bolus injections of 200,000 Cetus units each. In a long-term model, rabbits were treated with 3 (x) 10(6) Cetus units/kg of rIL-2 daily in three divided doses every 8 h for 9-11 doses (a schedule analogous to one used in human trials). Following treatment, animals were killed and examined for evidence of pulmonary toxicity. Among the treatment and control groups, there was no evidence of pulmonary leukostasis as determined by mean alveolar septal wall granulocytes per high power field or mean lung myeloperoxidase activity per gram of tissue. While there were no differences among the three treatment groups with regard to pulmonary edema formation (wet/dry weights), there was a tendency toward statistically significant differences between the rIL-2 and control groups. Pulmonary vascular permeability was assessed using i.v.-administered [125I]rabbit serum albumin (RSA) and expressed as mean bronchoalveolar lavage fluid/plasma [125I]RSA ratios. The rIL-2-treated animals had significantly increased pulmonary extravasation of [125I]RSA compared to controls, but there were no differences between the excipient- and D5W-treated controls. Lungs from rIL-2-treated (but not controls) animals displayed marked ultrastructural changes by electron microscopy in the arteriolar segment to include intracellular and subcellular blebs throughout the arteriole with separation of endothelial cells from basement membrane, interstitial edema, endothelial adhesion, and transmigration of lymphocytes into interstitium. Immunoperoxidase stains using an antirabbit T-cell monoclonal antibody demonstrated significant T-cell infiltration into the pulmonary interstitium of rIL-2-treated animals compared to controls. The long-term treatment model described appears to be highly suited for studies of rIL-2-induced pulmonary toxicity.