Activated cryptic granulocyte-macrophage colony-stimulating factor autoantibodies in intravenous immunoglobulin preparations

To the editor:

In their recent paper,¹  Uchida and collaborators discussed the frequent presence of granulocyte-macrophage colony-stimulating factor (GM-CSF) autoantibodies in pharmaceutical intravenous immunoglobulin (IVIg) preparations purified from the plasma of healthy donors, although these autoantibodies were previously reported to be rarely present in the plasma of healthy persons.²  They proposed, as an explanation for this apparent discrepancy, that GM-CSF autoantibodies were normally present in the plasma of healthy subjects at low levels and neutralized by binding to GM-CSF. This conclusion was derived from the results of experiments using low acidic pH conditions (pH 2.8) to dissociate GM-CSF from their autoreactive immunoglobulin G (IgG) followed by physical separation of the IgG and GM-CSF. This procedure was based on work by Watanabe and collaborators who used a similar protocol to show the presence of GM-CSF autoantibodies bound to GM-CSF in the plasma of healthy persons.³  The reported concentration of GM-CSF eluted from these putative immune complexes was extremely low (ie, below the sensitivity threshold of the ELISA used), yet Watanabe et al concluded that GM-CSF (approximately 10 pg/mL) was present in the eluates. This very low level of GM-CSF is much lower than the relatively high concentration of GM-CSF autoantibodies reported by Uchida et al (∼ 1 μg/mL). Such a difference does not support the hypothesis that GM-CSF autoantibodies are in a nonreactive state in the plasma of healthy persons due to the binding of their soluble antigen (GM-CSF).

Our recent work provides an alternative explanation for the presence of GM-CSF autoantibodies in IVIg preparation.⁴  We demonstrated that autoreactive GM-CSF antibodies were barely detectable in IgG prepared from small pools of plasma obtained from healthy persons using nondenaturing conditions (native IgG), but that treatment of native IgG under slightly denaturing conditions (pH 3.5 and 3.2) resulted in a significant increase in GM-CSF autoreactivity. In our experiments, the pH-treated native IgG preparations were neutralized without physical separation from the antigens. This would have allowed the reformation of immune complexes and the inhibition of the autoreactive IgG if GM-CSF had been present in the preparations. The pH condition (pH 2.8) used by Uchida et al is even more denaturing than those used in our study and has been known to efficiently activate cryptic autoreactive IgG.⁵  Therefore, we propose that autoreactive GM-CSF immunoglobulins are present in a cryptic state rather than in the form of immune complexes in the plasma of healthy persons and are detected in IVIg preparations as a consequence of modifications to the IgG structure induced by the industrial fractionation process. This conclusion could be extended to other previously described cytokine-reactive IgG observed in serum of healthy persons after low-pH exposure³  and in IVIg preparations.^2,6