Direct quantification of exhausted T cells in human cancer is lacking, and its predictive value for checkpoint-based treatment remains poorly investigated. We sought to systematically characterize the pan-cancer landscape and molecular hallmarks of T-cell dysfunction for the purpose of precision immunotherapy. Here, we defined a transcriptional signature for T-cell exhaustion through analyzing differential gene expression between PD-1-high and PD-1-negative CD8+ T lymphocytes from primary non-small cell lung cancer (NSCLC), followed by positive correlation tests with PDCD1 in TCGA lung carcinomas. A 78-gene signature for exhausted CD8+ T cells (GET) was identified and validated to reflect dysfunctional immune state spanning different species and disease models. We discovered that GET estimation significantly correlated with intratumoral immune cytolytic activity (CYT) and T-cell-inflamed gene expression profile (GEP) across 30 solid tumor types. Miscellaneous tumor-intrinsic and -extrinsic properties, in particular leukocyte proportions, genomic abnormalities, specific mutational signatures, and signaling pathways, were notably associated with GET levels. Furthermore, higher GET expression predicted an increased likelihood of clinical response to immune checkpoint inhibitors. These findings highlight the interrelation between T-cell exhaustion and immune cytolytic activity at the pan-cancer scale. The resulting inflamed tumor microenvironment may further crosstalk with other molecular and clinicopathological factors, which should be properly considered during immunotherapy biomarker development. © 2020 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.
Keywords: T-cell exhaustion; immune checkpoint inhibitors; immune cytolytic activity; inflamed tumor microenvironment; pan-cancer analysis.
© 2020 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.