Abstract
INTRODUCTION: Despite the lack of efficacy with immune checkpoint inhibitors (ICIs) in glioblastoma, there is evidence that neoadjuvant ICI impacts the glioblastoma tumor microenvironment and promotes survival in subsets of patients. Given the non-redundant mechanisms of T cell dysfunction mediated by different checkpoints, there is growing interest in combining ICIs for added efficacy. However, the immune responses of patients receiving dual ICIs, specifically anti-PD-1 and anti-LAG-3, remain poorly understood. Objective: We aimed to determine the impact of combination ICIs on glioblastoma T-cell landscape to elucidate determinants of immunotherapy response and resistance. METHODS: We performed single-cell RNA and paired TCR sequencing on tumor infiltrating lymphocytes from 27 IDH-wildtype glioblastoma patients (22 treatment-naïve; 5 treated with dual ICIs). RESULTS: We found that the most clonally expanded population of T cells in glioblastoma consisted of developmentally plastic GZMK+ T cells that are capable of diverging into three distinct states: GZMK+ resident T cells, terminal effector/dysfunctional T cells, and T cells re-expressing CD45RA. Our results revealed a paucity of putative tumor-reactive T cell clones in untreated glioblastoma patients. Treatment with dual ICIs recruits novel tumor-reactive T cell clones from the periphery and drives GZMK+ T cells toward an effector and dysfunctional trajectory along an exhaustion gradient characterized by upregulated TCR signaling, antigen presentation, proliferation, and multiple checkpoints. CONCLUSIONS: Our work identifies a population of GZMK+ effector memory T cells as the primary target of dual ICIs in glioblastoma with tumor-reactive T cells almost exclusively present in this transitional population. Dual ICI recruits novel T cell clones from the peripheral and preferentially activates GZMK+ T cells toward terminal effector and dysfunctional states. Our findings provide key mechanistic insights into how dual ICIs reshapes the T cell landscape and repertoire in glioblastoma and identify potential mechanisms of immunotherapy resistance, highlighting opportunities for therapeutic optimization.