Prediction of Aggregation In Vivo by Studies of Therapeutic Proteins in Human Plasma

Abstract

Proteins may aggregate after subcutaneous (s.c.) or intramuscular (i.m.) injection. This local in vivo aggregation is not intended or desired for most protein therapeutics. However, advantage has been taken of this phenomenon for formulation of slow-release products such as the long-acting insulin analog, Lantus^® or the peptidic gonadotropin-releasing hormone receptor blocker, Firmagon^®. Aggregation in vivo may also take place after intravenous (i.v.) or intra-arterial (i.a.) infusion of biopharmaceuticals, very likely increasing the risk of adverse events and enhancing immunogenicity. Evaluation of the potential to develop in vivo aggregation after administration of therapeutic proteins requires development of in vitro models to predict this aggregation. One model consists of mixing antibody formulations with human plasma at concentrations similar to the ones used in vivo. For example, mixing Herceptin^® and Avastin^® in 5 % dextrose alone failed to reveal aggregates. However, mixing these formulations with human plasma resulted in the formation of large aggregates. No such aggregation was observed when the antibodies were formulated in 0.9 % NaCl and mixed with human plasma. The aggregates which formed in the plasma–5 % dextrose–Herceptin^® and plasma–5 % dextrose–Avastin^® mixtures were analyzed by particle-flow imaging (PFI), nanoparticle tracking analysis (NTA) and electron microscopy (EM) methods. The observed rapid aggregation, the spherical shape of the aggregates and the electron microscopy structures suggest that the aggregates consist of antibodies and plasma components, probably lipoproteins. A model for the aggregation of antibodies in human plasma is proposed. These more advanced in vitro studies are critical in evaluation of biosimilars and biobetters to innovator products.