Predictive power of preclinical studies in animals for the immunogenicity of recombinant therapeutic proteins in humans.

Despite advances at the level of gene sequence (eg, humanized versus murine antibodies), expression systems (eg, mammalian versus prokaryotic) and post-expression modification (eg, 'PEGylation'), clinical immunogenicity of therapeutic proteins remains a concern. Although animals are routinely and effectively used to evaluate pharmacological activity and to support a claim of safety for recombinant proteins, their usefulness for predicting clinical immunogenicity is more questionable. This review argues that recombinant proteins can be grouped into immunogenic classes; for some of these classes, for example, bacterial proteins, immunogenicity in animals is often predictive for humans, but for others, for example, fully human proteins, even data from non-human primates can have little predictive value. We will attempt to make the case that for a variety of immunological and practical constraints, animal studies, even those conducted in non-human primates, have limited predictive power for immunogenicity in humans, and tend to over-predict clinical immunogenicity.