The pharmacokinetics of recombinant human granulocyte-macrophage colony-stimulating factor (rhGM-CSF), induction of anti-GM-CSF antibodies, and clinical effects related to the induction of the antibodies were analyzed in patients with metastatic colorectal carcinoma (CRC) who were not on chemotherapy (n = 20, nonimmunocompromised patients). rhGM-CSF (250 micrograms/m2/d; Escherichia coli-derived) was administered subcutaneously for 10 days every month for 4 months. Eight patients with multiple myeloma (MM) on intensive chemotherapy followed by rhGM-CSF treatment were also included (immunocompromised patients). After a single injection of GM-CSF at the first cycle in CRC patients, the maximum calculated concentration (Cmax) was 5.24 +/- 0.56 ng/mL; the half life (T1/2) was 2.91 +/- 0.8 hours; and the area under the concentration curve (AUC) was 30.86 +/- 6.03 hours x ng/mL (mean +/- SE). No anti-GM-CSF antibodies were detected. During the subsequent cycles, 95% of the CRC patients developed anti-GM-CSF IgG antibodies, which significantly altered the pharmacokinetics of rhGM-CSF at the third and fourth cycles with decreased Cmax (2.87 +/- 0.57 ng/mL; P < .05), T1/2 (1.57 +/- 0.2 hours; P < .05), and AUC (14.90 +/- 4.10 hours x ng/mL; P < .005). The presence of anti-GM-CSF antibodies significantly reduced the GM-CSF-induced enhancement of granulocytes, and there was a clear tendency for a decreased increment of monocytes. Antibodies diminished systemic side effects of rhGM-CSF. Only 1 of 8 MM patients showed a very low anti-GM-CSF antibody titer after GM-CSF therapy, as shown by enzyme-linked immunosorbent assay and Western blot. Therefore, in nonimmunocompromised patients, exogenous nonglycosylated GM-CSF induced an anti-GM-CSF IgG antibody response in practically all patients, which seemed to be of clinical significance. In immunocompromised patients, virtually no significant antibody response was shown.