A predictive biomimetic model of cytokine release induced by TGN1412 and other therapeutic monoclonal antibodies.

Human peripheral blood mononuclear cells (PBMC) are routinely used in vitro to detect cytokine secretion as part of preclinical screens to delineate agonistic and antagonistic action of therapeutic monoclonal antibodies (mAbs). Preclinical value of standard human PBMC assays to detect cytokine release syndrome (CRS) has been questioned, as they did not predict the "cytokine storm" that occurred when healthy human volunteers were given a CD28-specific super-agonist mAb, TGN1412. In this article, we describe a three-dimensional biomimetic vascular test-bed that can be used as a more physiologically relevant assay for testing therapeutic Abs. For developing such a system, we used TGN1412 as a model mAb. We tested soluble TGN1412 on various combinations of human blood components in a module containing endothelial cells grown on a collagen scaffold and measured cytokine release using multiplex array. Our system, consisting of whole leukocytes, endothelial cells, and 100% autologous platelet-poor plasma (PPP) consistently produced proinflammatory cytokines in response to soluble TGN1412. In addition, other mAb therapeutics known to induce CRS or first infusion reactions, such as OKT3, Campath-1H, or Herceptin, generated cytokine profiles in our model system consistent with their in vivo responses. As a negative control we tested the non-CRS mAbs Avastin and Remicade and found little difference between these mAbs and the placebo control. Our data indicate that this novel assay may have preclinical value for predicting the potential of CRS for mAb therapeutics.