Anti-Gal antibodies—where's the beef?

To the editor:

In your July issue¹, Carl Borrebaeck avers that murine glycosylation patterns on murine antibodies, and on human antibodies produced in murine cells, will target them for rapid clearance by natural anti-Gal antibodies in humans. But the data defy him.

First there is no evidence for reactivity in vivo. The detection of weakly reactive antibodies in an ELISA assay¹ is not the most appropriate model for what happens in vivo. Many polyreactive antibodies of low affinity exist in serum², but at concentrations is generally well below their K_d; little of the antigen will be bound in complex. Moreover, use of purified preparations for these assays is known to overstate significantly the effective immunotargeting of polyreactivities in serum³. If anti-Gal antibody were a real phenomenon affecting survival of murine or murine-produced antibodies, it would routinely reveal itself as high levels of HAMAs (or HAHAs) in baseline serum assays in patients, but this is not the case.

Second, there is no physical evidence of such complexes in vivo. We have examined dozens of samples of tracer- or therapeutically radiolabeled mouse and human(ized) antibodies on HPLC sizing columns, either drawn from patients after infusion or by spiking labeled antibody into sera. All show a homogeneous peak in human sera, which is not displaced relative to the same antibody incubated in saline⁴. Yet soluble antigen–antibody complexes are readily detectable in individuals with high antigen levels^4,5,6,7, in a relation^4,5,7 that is not to be confused with the postulated anti-Gal antibody–antibody complexes. If there is an immune recognition of significance against Gal-containing mouse or human antibodies, it should be routinely detectable as complexes in vivo at baseline, but this is not the case.

Finally, an abbreviated half-life due to anti-Gal reactivity is simply not true. The humanized anti-Tac antibody, Zenapax, is grown in mouse cells with murine glycosylation patterns. In 92 patients treated, Zenapax had an average survival time of 20 days¹⁰, essentially indistinguishable from "human-grown" human IgGs (23 ± 4 days)¹¹. Similarly for Synnapsis¹² ans ABX-IL8¹³. Thus of three humanized or human antibodies grown in mouse cells for which clinical data are available, none exhibits an accelerated clearance in humans.

There is no evidence for a baseline anti-antibody-mediated clearance, by anti-Gal or any other humoral activity. Accordingly, any further debate should be accompanied by new, clinical data that directly addresses the hypothesis, which has so far been lacking.