Over two, well attended sessions in September 2021, the topic of cytokine release assays (CRA) for hazard prediction was discussed. The goal of these sessions was to introduce consortium members to the classical CRA formats and how they are currently used in drug development and to provide an opportunity for partners to share their experiences with novel in vitro assay formats. These novel formats have been designed to more closely recapitulate the mechanisms of clinical cytokine release syndrome and could be used to improve hazard prediction and dose prediction for first-in-human clinical trials.
Introduction: The current in vitro cytokine release assay formats were developed in response to the experiences with TGN1412, a monoclonal anti-CD28 superagonist antibody, where severe cytokine release syndrome (CRS) was observed in the First-in-Human clinical study that was not predicted by the conducted pre-clinical studies. Current formats are now widely used as an important screening tool for immunomodulatory therapeutics; however, gaps remain in their efficient and translatable quantification of clinical CRS risk. The most commonly used formats use whole blood or peripheral blood mononuclear cells (PBMCs) with either soluble or immobilized drug but can be adapted based on the mechanism of action of the tested molecules. Currently there is no one-size-fits-all approach for selecting the most appropriate assay format for a specific molecule, and regulatory and technical best practices are constantly evolving.
BOEC assay system: The blood outgrowth endothelial cells (BOEC) in vitro cytokine release assay is a novel assay developed by Jane Mitchell’s group at Imperial College London and now implemented at Labcorp, Harrogate. The assay is a fully autologous co-culture of PBMCs or BOEC isolated and expanded from the same human donor. The purpose of developing such a system was to reflect the physiological microenvironment of the blood vessel as IL-6 released from vascular endothelial cells plays a critical role in amplifying clinical cytokine storm events. The benefit of using autologous endothelial cells was to reduce any background noise due to allogeneic reactions. Data showed a strong release of cytokines in response to TGN1412 whereas molecules with limited clinical CRS were negative in the assay. The assay was also increased in sensitivity compared to classical PBMC only based formats.
MIMIC system: TheModular Immune In-vitro Construct (MIMIC) system is an in vitro platform that was initially developed to evaluate vaccine immunogenicity. The innate immune system module forms the basis of this new in vitro cytokine release assay. Endothelial cells are seeded into plates in a collagen matrix and incubated with erythrocyte depleted whole blood and media supplemented with autologous plasma. In this assay format, phenotyping of PBMCs by flow cytometry is also feasible in addition to the analysis of cytokines released. In response to stimulation with TGN1412 cytokine release can be detected and the platform was considered more sensitive than the classical PBMC based assay. In addition, the MIMIC system was used to assess the cytokine release upon stimulation with a T cell engaging bispecific molecule. Here, it could be shown that the molecule led to a clear Th1 cytokine response that was aligned with the T cell activation status as measured by flow cytometry (frequency of CD4+CD25+ and CD8+CD25+ cells). A dose-dependent reduction in donor B cells was observed and aligned with expected pharmacology at a functional level using flow cytometry. Furthermore, in first test runs using CD19-specific CAR-T cells as the therapeutic agent, the cytokine response measured in the MIMIC assay showed good correlation with cytokine release patterns observed in clinical studies and it was also used to demonstrate the requirement of endothelial cells to mediate the strong cytokine release.
Use of cytokine release assays in industry: A selection of case examples where cytokine release assays have been used during the drug development process were presented. Case 1: a monoclonal antibody developed for a hematological oncology indication whose mechanism of action was predominantly antibody-dependent cellular cytotoxicity (ADCC). Cytokine release assays were conducted in whole blood with both soluble and plate-bound drug and these formats were selected to include the relevant target (granulocyte and monocyte) populations and to allow for crosslinking and Fc-gamma receptor engagement. In response to a question from a regulatory agency to further determine the MoA of the observed cytokine release in a more sensitive model, an optimized in vitro assay (high-density PBMC culture) was conducted. It was concluded that the initially conducted whole blood assays were the most sensitive model due to presence of target cells, which were underrepresented in PBMC preparations. Case 2: a checkpoint inhibitor molecule where cytokine release was expected based on the mechanism of action. In this case, cytokine release was assessed in parallel to the pharmacology in a PBMC based assay with antigen stimulation. Case 3: a T cell engaging bispecific molecule where cytokine release was an expected consequence of the mechanism of action. In line with recent health authority guidance, no specific cytokine release assays were performed as cytokine release could be assessed in the pharmacological assays using target cell and PBMC co-culture.
Vessel-on-a-chip: An actively perfused blood vessel on a chip was created using a controlled, engineered microenvironment. The chip included a vessel wall created from endothelial cells (HUVEC or hiPSC-EC) and was perfused with human whole blood. This model was used to assess the effect of drugs on thrombus formation including GPIIb/IIIa inhibitors and anti-platelet medications. In a case example, an anti-CD40L antibody in the presence of its ligand sCD40L caused an Fc-gamma Receptor-dependent increase in thrombus formation. The model can also be modified with increasing complexity to address specific questions e.g. to assess epithelial-endothelial cell crosstalk by co-culturing with primary alveolar epithelium or cardiac cell crosstalk using a co-culture of iPSC cardiac tissue with vascular endothelium. Finally, a vessel-on-chip model with a 3D structure in a collagen matrix is also under development.
Vessel-on-a-chip for cytokine release assays: A 3D vessel-on-a-chip using BOEC cells which can be perfused with either whole blood or PBMCs was developed to improve the assessment of cytokine release. The system aimed to increase the physiological relevance by incorporating flow conditions into the system. In collaboration with Mimetas using the Organoplate 2-Lane platform a 3D vessel-like structure was formed with the BOEC cells. Cytokine release assays were performed by the addition of whole blood or PBMCs, rocking to provide bi-directional flow and test compound stimulation in soluble form. Initial data with test compounds was promising with the potential to identify cytokine release in the 3D system that would not have been identified in the 2D system for certain compounds.
We thank Hannah Morgan, Chris Cooper, Ernesto Luna, Birgit Fogal, and Andries van der Meer for their presentations.