Signatures of immune senescence predict outcomes and define checkpoint blockade-unresponsive microenvironments in acute myeloid leukemia

Summary The function of senescent-like T cells, transcriptomic features of immune effector senescence (IES) and their influence on therapeutic response were investigated in independent AML clinical cohorts comprising 1,864 patients treated with chemotherapy and/or immune checkpoint blockade (ICB). We show that senescent-like bone marrow CD8 + T cells are impaired in killing autologous AML blasts, and that their proportion negatively correlates with overall survival (OS). We define new IES signatures using two gene expression platforms and report that IES scores correlate with adverse-risk molecular lesions, stemness, and poor outcomes as a potentially more powerful predictor of OS than 2017-ELN risk or LSC17 stemness score. IES expression signatures also identify an ICB- unresponsive tumor microenvironment and predict significantly worse OS in AML as well as in solid tumors. The newly described IES scores provide improved AML risk stratification and could facilitate the delivery of personalized immunotherapies to patients who are most likely to benefit.