In vitro cytokine release assays for commercially available biologics for predicting cytokine storm in human patients (P6321)

Abstract Therapeutic antibodies undergo numerous safety tests to ensure in vivo safety. One required safety assay looks at the cytokine release from either human whole blood or human peripheral blood mononuclear cells (PBMC) when incubated with the therapeutic antibody. Recently TGN1412, a CD28 super-agonist antibody, while not causing cytokine release from purified PBMC, when administered in vivo, resulted in cytokine storm in all six volunteers. Since that incident, many laboratories have been working to design a cytokine release assay (CRA) that will show cytokine release in response to TGN1412. In this study we compare two CRA formats. In one assay, test antibody is coated to a polypropylene plate, then incubated with freshly purified PBMC for two days. In the second assay, human PBMC are pre-incubated at high density to make the cells more "lymph node like" (LNL) before incubation with soluble antibody. The response of PBMC from multiple donors to TGN1412 and multiple commercially available antibodies were compared. The plastic fixed antibody tends to generate exaggerated cytokine responses probably due to excessive crosslinking. The LNL format has the advantage of using soluble antibody which may better reflect how PBMC in vivo encounter antibody.