Combined Human and In Vivo Analyses Implicate the Critical Importance of Acute Myeloid Leukemia Associated CD8 T Cells and Identifies Their Molecular Dysfunction

Successful immunotherapies for acute myeloid leukemia (AML) remain elusive, partly due to the incomplete characterization of leukemia-induced changes to the T cell landscape. Here, we present proteomic, transcriptomic, and functional studies of the AML T cell landscape using patient samples, two immunocompetent (radiation-free) mouse models of leukemia, and publicly available datasets.

We immunophenotyped cells from healthy donor marrow (n=5) and AML patient samples (n=35) using mass cytometry with 20-plex mass-tag barcoding. We used a 38-marker panel to identify progenitor, myeloid, and lymphoid populations. In parallel, we used a 37-marker panel with a focus on T cell phenotypes to identify their subpopulations. Computational debarcoding, batch correction, dimensional reduction, and clustering were followed by expert identification of cell types based on canonical surface marker expression. Within the CD8 T cell compartment, AML samples had significantly less abundant naïve T cells and more abundant terminal effector T cells compared to healthy controls. The terminal effector CD8 T cell cluster was characterized by expression of CD57, perforin, and granzyme B.

AML cells are also immune cells and may have unique immunomodulatory properties. Therefore, we hypothesized that CD8 T cells in AML are transcriptionally distinct from CD8 T cells from solid tumors. We integrated publicly available scRNAseq data from AML (Desai et al, 2023) and pan-cancer tumor infiltrating lymphocytes (TILS, Zheng et al, 2021). Differential expression analysis comparing CD8⁺ T cells in AML to CD8⁺ TILS in breast cancer, esophageal cancer, thyroid cancer, and endometrial cancer revealed relative upregulation of KLRG1, IFNG, and TNF in AML-associated CD8 T cells.

Next, we tested the anti-leukemic role of CD8 T cells in an immunocompetent mouse model of AML. C57BL/6J-Ptprc^em6Lutzy/J CD45.1⁺ host mice were transplanted with CD45.2⁺ leukemic cells bearing the MLL-AF9 translocation and then treated with αCD8α depleting antibody or isotype control. Mice treated with the αCD8α depleting antibody trended toward shorter survival (median, 23.0 days αCD8α vs 26.5 days control), albeit was not significant (log rank test, p=0.076). We hypothesized that AML-associated CD8 T cells develop a dysfunctional phenotype that precludes them from exerting their effector functions. Indeed, within the CD8 T cell compartment, mice with MLL-AF9 leukemia had significantly more abundant PD1⁺, LAG3⁺, and TIM3⁺ cells compared to those transplanted with healthy, wild-type marrow. These differences in CD8 T cell phenotypes were replicated in a second immunocompetent mouse model, wherein C57BL/6J-Ptprc^em6Lutzy/J host mice were transplanted with leukemic cells bearing FLT3^ITD-NPM1^c mutations.

We then investigated if AML-associated CD8 T cells could exert their effector functions ex vivo. After phorbol myristate acetate and ionomycin stimulation, a significantly higher proportion of CD8 T cells from MLL-AF9 produced TNFα and IFNγ compared to their healthy counterparts by flow cytometry. Finally, we conducted RNA sequencing of the CD44⁺CD62L^- (effector) CD8 T cell population from MLL-AF9 marrow and healthy marrow. Gene set enrichment analysis of Hallmark pathways revealed significant enrichment of proliferation-related pathways such as E2F targets (NES=1.64), G2M checkpoint (NES=2.86), and mitotic spindle (NES=1.64) in AML-associated T cells in addition to the IL2 signaling pathway (NES=1.84). However, differential expression analysis showed increased expression of Tigit, Pdcd1, Havcr2, and Tox.

Here, we present several orthogonal studies to investigate AML-associated T cells in human samples and mouse models. Our data demonstrate that T cells in human AML, identified by mass cytometry and scRNAseq, are enriched for terminal effector CD8 T cells and display a unique phenotype among cancer subtypes. We then establish two immunocompetent mouse models with unique driver genes, and further show similar T cell differentiation and dysfunction to human samples. Finally, we demonstrate that while these T cells retain, and even increase, their ability to produce cytokines with stimulation, by transcriptional analysis display signs of impending dysfunction. This exhaustive characterization of T cell dysfunction in AML will enable future studies into the exploitation of the immune system for AML-directed T cell therapy.

Levine:Anovia: Consultancy; Syndax: Consultancy; Zentalis: Membership on an entity's Board of Directors or advisory committees; Bakx Therapeutics: Membership on an entity's Board of Directors or advisory committees; Kurome: Membership on an entity's Board of Directors or advisory committees; Mana: Membership on an entity's Board of Directors or advisory committees; Scorpion: Membership on an entity's Board of Directors or advisory committees; C4 Therapeutics: Membership on an entity's Board of Directors or advisory committees; Jubilant: Membership on an entity's Board of Directors or advisory committees; Prelude Therapeutics: Membership on an entity's Board of Directors or advisory committees; Bridge Medicines: Consultancy; Bridge Bio: Consultancy; Isoplexis: Membership on an entity's Board of Directors or advisory committees; Imago: Consultancy; Epiphanes: Membership on an entity's Board of Directors or advisory committees; Auron: Membership on an entity's Board of Directors or advisory committees; Ajax: Membership on an entity's Board of Directors or advisory committees; Mission Bio: Membership on an entity's Board of Directors or advisory committees; Qiagen: Membership on an entity's Board of Directors or advisory committees. Ferrell:Novartis: Research Funding.