T cell engaging therapies are commonly accompanied by excessive cytokine production and risk of cytokine release syndrome (CRS). Intriguingly, CRS risk with CD3-engaging bispecific antibody (BSP) is primarily limited to the first dose, termed the first-dose effect. Mechanisms underlying this effect remain unknown. CD3 bispecific induces cytokine cascade via T cell triggering and bystander cells. We hypothesize that distinct T cell biology between doses drives the first-dose effect.

We used the Re-directed T Cell Cytotoxicity (RTCC) assay to assess tumor killing and cytokine production by human donor T cells after initial versus subsequent CD3/CD20 BSP treatment. After confirming the first-dose effect in the experimental system containing only T cells and target tumor cells, we employed 10x Genomics single cell multi-omics to study the molecular mechanisms.

Compared with initial CD3/CD20 BSP treatment, subsequent treatment exhibited lower cytokine levels and comparable tumor killing. Single cell multi-omics unveiled distinct T cell biology. In initial treatment, T effector memory (Tem) cells are the primary cells that respond to CD3 bispecific antibody stimulus by producing moderate levels of cytolytic and high levels of cytokine gene transcription. In the subsequent treatment, a new population of high TCF7 expressing central memory CD8 + cells (CD8-Tcm-TCF7), possibly originated from stimulated naive T cells, are the primary responding cells that produce a shifted balance with high level of the cytolytic gene transcription (GZMB) and low level of cytokine gene transcription (TNF-alpha and IFN-gamma). Dasatinib co-treatment during initial treatment eliminated cytolytic activity and cytokine production, allowing uncompromised tumor killing and reduced cytokine production upon re-challenge.

The distinct T cell populations that respond to first and subsequent CD3 bispecific treatment offer an explanation to the first-dose effect, wherein the risk of CRS associated with CD3 bispecific treatment is mainly limited to the initial dose. Furthermore, our work suggests that tumor killing capacity and cytokine production of T cells could be uncoupled, as demonstrated here by utilizing different T cell populations as effector cells. These findings could be further explored for designing mechanism-based strategies to mitigate the risk of CRS.