GM-CSF Blockade during Chimeric Antigen Receptor T-Cell (CART) Therapy Reduces Cytokine Release Syndrome and Neurotoxicity and May Enhance CART Effector Functions

Abstract

Chimeric antigen receptor T-cell therapy (CART) is limited by the development of cytokine release syndrome (CRS) and neurotoxicity (NT). CRS is related to extreme elevation of cytokines and massive T cell expansion. Preliminary studies suggest that NT might be mediated by myeloid cells that cross the blood brain barrier. This is supported by correlative analysis from CART19 pivotal trials where CD14+ cell numbers were increased in the cerebrospinal fluid of patients that developed severe NT (Locke et al, ASH 2017). Thus, we aimed to investigate the role of GM-CSF neutralization in preventing CRS and NT after CART cell therapy via monocyte control. The human GM-CSF neutralizing antibody lenzilumab (Humanigen, Burlingame, California), which has been shown to be safe in phase II clinical trials, does not impair CART cell function in vitro (10 ug/mL) when CART19 cells are stimulated with the CD19+ Luciferase+ acute lymphoblastic leukemia (ALL) cell line NALM6. Lenzilumab results in enhanced CART cell antigen specific proliferation in the presence of monocytes. In vivo, NOD-SCID-g−/− mice were engrafted with high disease burdens of NALM6 and treated with low doses of CART19 or control T cells, in combination with lenzilumab or isotype control antibody. The combination of CART19 and lenzilumab does not impair CART cell activity in vivo (Fig 1A). The impact of GM-CSF neutralization on CART toxicities in a novel patient derived xenograft model was explored. NOD-SCID-g−/− mice were engrafted with primary ALL blasts (1-3 × 106 cells). Mice were treated with high doses of CART19 cells (2-5 × 106 iv) and developed CART cell related toxicities. The combination of CART19, lenzilumab (to neutralize human GM-CSF), and murine GM-CSF blocking antibody (to neutralize mouse GM-CSF) resulted in prevention of weight loss (Fig 1B), decrease in critical myeloid cytokines (Fig 1C-D), reduction of cerebral edema (Fig 1E), enhanced leukemic disease control in the brain (Fig 1F), and reduction in brain macrophages (Fig 1G). As CRISPR/Cas9 gene editing of the GM-CSF gene during the process of CART cell manufacturing would result in CART cells with innately reduced secretion of GM-CSF, we designed guide RNA targeting exon 3 of the GM-CSF gene to generate GM-CSFk/o CART19 cells. These CARTs produce significantly less GM-CSF upon activation but other T cell cytokine production is not inhibited (Fig 1H). Using the NALM6 high tumor burden relapse xenograft model as described above, GM-CSFk/o CART19 cells resulted in slightly enhanced disease control compared to CART19 cells (Fig 1I). These studies illuminate a novel approach to abrogate NT and CRS through GM-CSF blockade that also potentially enhances CART cell functions. Based on these results, we have designed a phase II clinical trial using lenzilumab as a modality to prevent CART related toxicities in patients with diffuse large B cell lymphoma.