Pre-Existing Anti-GM-CSF Autoantibodies in Patients with AML, CML and MDS Are Associated with Compromised Immune Response to PR1 Peptide Vaccination.

Recombinant Human Granulocyte-Macrophage Colony-Stimulating Factor (rhGM-CSF) is used as an adjuvant in tumor vaccination to boost the immune response by causing outgrowth and maturation of dendritic cells. Autoantibodies to GM-CSF that block its activity have been described in patients with alveolar proteinosis, and administration of pharmacological doses of rhGM-CSF might result in the induction of anti-GM-CSF antibodies, especially when rhGM-CSF is injected with other adjuvants. Our laboratory has identified PR1, an HLA-A2-restricted 9 amino acid peptide derived from proteinase 3, as a leukemia-associated antigen for leukemia-reactive cytotoxic T lymphocytes (CTL). In a phase I/II clinical trial, patients with myeloid leukemia received PR1 peptide as a subcutaneous vaccine in incomplete Freund’s adjuvant (IFA) with a subsequent injection of 75mg rhGM-CSF every 3 weeks for 3 total injections. Immune response to vaccine was evaluated by PR1/HLA-A2 tetramer staining and clinical response was evaluated by NCI criteria, as have been reported previously (ASH, 2004). We sought to determine whether the PR1 vaccine induced GM-CSF-neutralizing antibodies, based on the hypothesis that such antibodies may compromise the adjuvant effect. Using ELISA, Western blot and a TF-1 cells-based GM-CSF neutralization assay, we tested 9 AML, 5 MDS and 5 CML patients that received PR1 vaccine for anti-GM-CSF antibodies. We identified anti-GM-CSF IgG in 6 out of 19 (32%) patients (53±17RU/ml) compared to none out of 33 (6±2RU/ml) healthy donors (p=0.008). Additionally we detected anti-GM-CSF IgM and IgA in 5 (33%) and 3 (20%) out of 15 patients, respectively. To our surprise, anti-GM-CSF IgG, IgA and IgM titers did not increase after vaccination (p=0.18; N=17) suggesting that the antibodies were present before treatment. Moreover, pre-existing anti-GM-CSF IgG titers were higher (113±40RU/ml; n=8) in patients that did not have immune response after vaccine than in immune responders (37±26.1RU/ml; n=8)(p=0.05). Furthermore, none of six patients with higher anti-GM-CSF IgG had clinical remission after vaccination. In contrast, 6 out of 11 patients (54%) with lower anti-GM-CSF IgG had clinical response to vaccination (p=0.02). Additional analysis of 16 AML, 17 CML, 2 ALL and 4 CLL patients showed that anti-GM-CSF IgG were present only in patients with myeloid leukemia, compared to patients in remission (p=0.001) or those with lymphoid leukemia (p=0.001). Only sera from one patient studied showed neutralizing activity. Anti-GM-CSF IgG titers did not correlate with GM-CSF expression in peripheral blood cells (mRNA transcripts) or serum (protein), which were not different from healthy controls. Our data show that vaccination with PR1 peptide did not induce anti-GM-CSF antibodies. However, pre-existing autoantibodies were associated with lower rates of immune and clinical responses suggesting that anti-GM-CSF antibodies might compromise effective vaccination. As reported elsewhere, immune and clinical responses to PR1 vaccination correlate inversely with disease burden before vaccination. Here we demonstrate the association of anti-GM-CSF antibodies and presence of disease. Since anti-GM-CSF antibodies were mostly non-neutralizing we have not established physiological relevance of these autoantibodies. Nevertheless, this association needs to be explored further since this is also the first report of naturally occurring IgA and IgM anti-GM-CSF autoantibodies.