Immunogenicity of granulocyte-macrophage colony-stimulating factor (GM-CSF) products in patients undergoing combination therapy with GM-CSF

In this study, we have assessed the development of neutralizing and nonneut ralizing granulocyte-macrophage colony-stimulating factor (GM-CSF) antibodi es in two groups of patients with metastatic colorectal carcinoma receiving two different GM-CSF products. Three clinical trials were carried out, and a combination of GM-CSF and a colon carcinoma-reactive antibody was used i n the absence of any concomitant chemotherapy. Two different GM-CSF product s, both rDNA-derived and produced in Escherichia coil, were used. Patients in Trial 1 received product X, and those in Trials 2 and 3 received product Y. Patients in Trial 2 also received interleukin 2 in an attempt to potent iate immune responses. After the first cycle of treatment, no GM-CSF antibo dies were detected, but on subsequent therapy, 28 of the 38 patients tested receiving product Y (Trials 2 and 3) developed antibodies that bound to th e GM-CSF product used for therapy. However, none of the patients developed antibodies that neutralized the biological activity of GMCSF, as assessed u sing an in vitro bioassay. Furthermore, there was no in vivo impairment in GM-CSF-induced expansion of leukocytes, neutrophils, and eosinophils in the patients. In contrast, 19 of the 20 patients given product X (Trial 1) dev eloped GM-CSF binding antibodies, and 9 of these patients were shown to dev elop antibodies that neutralized the biological activity of GM-CSF. The pre sence of the latter,vas associated with a significant reduction in GM-CSF-i nduced expansion of leukocytes, neutrophils, and eosinophils in patients. T herefore, product X appears to be more immunogenic than product Y. Immunoch emical characterization confirmed that the specificity of the antibody resp onses varied depending on the product used for therapy. Whereas sera from T rial 1 patients treated with product X showed the presence of antibodies wi th strong recognition of GM-CSF proteins, sera from patients treated with p roduct Y showed varied recognition of GM-CSF ranging from fairly strong to very weak but bound predominantly to two E. coil-derived, non-GM-CSF-relate d proteins of M-r similar to 20,000 and M-r similar to 30,000. Therefore, i n sera from patients receiving product Y, the antibody specificity appeared to be directed not only against GM-CSF but also against nonproduct-related host cell contaminants. This study shows that GM-CSF products used for the rapy are potentially immunogenic and generate antibodies to GM-CSF and/or o ther non-product-related contaminants. However, only antibodies that neutra lize the biological activity of GM-CSF compromise therapeutic efficacy of t he cytokine.