Mechanistic Studies of Cytokine Release Syndrome (CRS) with Roles of Interferon-Gamma (IFN-g) and Tumor Necrosis Factor Alpha (TNF-α) While Maintaining CAR-T Function in Vitro

Background: Chimeric Antigen Receptor (CAR) T cell therapy in patients with hematologic malignancies is limited by immune-related toxicities such as Cytokine Release Syndrome (CRS) in over 30% of patients, whereas no CRS was observed in patients receiving NK cell therapy. CRS is characterized by an early phase with fever, hypotension, and elevations of cytokines, including IFN-γ, GM-CSF, TNF-α , IL-10, and IL-6.

Methods: To study the mechanisms of CRS, we developed a model of in vitro CRS assay in which different effector cells, tumor target cells, and immature dendritic cells (iDC) were co-cultured, and the induction of inflammatory cytokines was measured by ELISA. To generate immature dendritic cells (iDC) for the assay, monocytes were cultured in the presence of GM-CSF and IL-4 for 4 to 6 days with the addition of fresh cytokines every 2 or 3 days, and the assay was set up on day 5 of iDC culture. This “priming” step was essential for the production of IL-6 by iDC (confirmed by scRNA-seq) after exposure of CART to target cells. Effector cells including CAR-T cells (targeting CD19 or BCMA), CAR- Memory-Like NK cells (CAR-ML-NK), and CAR invariant NKT (CAR-iNKT) cells were evaluated in our model.

Results: In our initial in vitro CRS assay, we conducted an experiment by using CART19 (19-28BBζ, 25,000 cells), which was co-cultured with CD19+ tumor (Ramos^CBR/GFP, 50,000 cells) and iDC (5000 cells) for 24 to 48hrs. Secreted IL-6, a surrogate marker of CRS, was determined using a human IL-6 ELISA. In vitro killing efficacy of CART19 was determined by bioluminescent imaging (BLI) 24 hrs post co-culture of CART19 with Ramos^CBR/GFP targets. A marked induction of CRS-related cytokines (IL-6, IFNγ, TNFα) was observed 24 and 48 hr post-coculture. Similar results were also found when we used CAR-T cell targeting BCMA with BCMA+ MM.1S target and iDC. In contrast, neglibile levels of IL-6, IFNγ, and TNFα secretion were detected when BCMA CAR Memory-Like Natural Killer cells (MLNK) were used compared to BCMA CAR T cells in spite of comparable levels of in vitro killing of BCMA+ MM.1S target cells by both BCMA-CART and BCMA-CAR-MLNK. With the BCMA CAR-iNKT, levels of IL-6, IFN-γ, and TNF-α were intermediate to those of BCMA CAR T cells and BCMA-CAR-MLNK again in spite of identical killing of BCMA+ targets in vitro.

To test the effect of kinase inhibitors (Amatya et al, Blood,140:12686) and neutralizing antibodies to block CRS (IL-6 secretion by iDCs) and their impact on CAR-T cell killing, we performed the CRS assay in the presence or absence of small JAK1/2 inhibitors and PI3Kg/d inhibitors (ruxolitinib and duvelisib) and neutralizing antibodies (against IFN-γ, TNF-α, GM-CSF). Both ruxoliltinib and duvelisib and neutralizing antibodies (anti-IFN-γ and anti-TNF-α but not GM-CSF) significantly reduced IL-6 levels (Fig 1). In contrast to dasatinib which completely blocked both CRS and CART killilng (data not shown), no significant attenuation of CAR-T killing efficacy was found with any of these kinase inhibitors or the neutralizing antibodies. The CART CRS assay was adapted to measure both CRS (IL-6 secretion) and target cell killing in response to BiTEs such as blinatumomab. Similar to CART19 blinatumomab induced a dose-dependent increase in IL-6 secretion in the presence of T cells, target Ramos cells, and iDC. CRS (in vitro production of IL-6) was dependent on the presence of GM-CSF/IL-4 “primed” iDC and abolished by duvelisib while maintaining blinatumomab-induced Ramos killing (Fig 2). Finally, we also developed and in vitro mouse CRS assay by coculturing mouse CART19, target (A20^CBR/GFP) cells and mouse bone marrow-derived macrophages. Marked increase in secreted mouse IL-6 was observed which was completely inhibited by duvelisib, ruxolitinib, anti-IFNγ and anti-TNFα.

Conclusion: We have developed in vitro CRS co-culture models using cells from both man and mouse and tested the role of multiple effector cell types including CART cells, CARMLNK cells, and CAR-iNK cells and multiple inhibitors on CRS and CART function in vitro. We show that JAK1/2 and PI3Kg/d kinase inhibitors as well as neutralizing antibodies to IFNγ and TNFα all block in vitro CRS without attenuating the anti-tumor efficacy of CAR-T cells. Our result suggests IFNγ, TNFα, JAK1/2 and PI3Kg/d kinases inhibitors are rational targets for CRS mitigation approaches without compromising CART-mediated anti-tumor efficacy in the clinic.