CD39 defines an exhausted effector memory T cell subset associated with poor outcome in Acute Myeloid Leukemia: Immunophenotypic and transcriptomic analysis

초록

Recent advances in cancer immunology have highlighted the role of T cell exhaustion and immune checkpoint pathways in shaping the tumor microenvironment of various types of cancers. The immunological landscape of T cell subsets in acute myeloid leukemia (AML) remains incompletely understood. CD39, an ectonucleotidase expressed on activated and exhausted T cells, is emerging as a critical marker of immunoregulatory function and dysfunctional anti-tumor immunity. Herein, we investigated the immunological significance and clinical relevance of CD39 expression on T cell subsets in a prospective cohort of newly diagnosed AML patients treated with standard 7+3 induction chemotherapy. Using multiparametric flow cytometry, we first analyzed CD39 expression in bone marrow T cells at diagnosis. t-SNE visualization revealed distinct clustering of CD4+ and CD8+ populations, with a notable enrichment of CD39+ cells in the effector memory T cell (TEM, CD45RA-CCR7-) and CD45RA-expressing effector memory T cell (TEMRA, CD45RA+CCR7-) compartments in both subsets. Quantitative analysis showed a stepwise increase in CD39 expression along the T cell differentiation axis, peaking in the TEM population, suggesting a potential link between CD39 expression and differentiation- or exhaustion-related states. Given that CD39 is known to be expressed in CD4+ FoxP3+ regulatory T (Treg) cells, we further examined its distribution across Treg and non-Treg CD4⁺ T cells. While CD39⁺ cells were enriched among Tregs cell population (CD25hiCD127lo/-FoxP3+), a greater proportion of CD39⁺ CD4+ T cells were found in the non-Treg compartment. Notably, the frequency of CD39⁺ cells was higher in non-Treg CD4⁺ T cells than in Treg cells. CD39⁺ non-Treg CD4⁺ T cells exhibited significantly higher expression of PD-1 and TIGIT compared to their CD39⁻ counterparts. This suggests that the immunosuppressive role of CD39 within the CD4⁺ T cell compartment may not be solely attributed to classical Tregs but may also reflect an independent and active role of CD39⁺ non-Treg CD4⁺ T cells. To further investigate the functional implications of CD39 expression, we performed bulk RNA sequencing on sorted CD39⁺ and CD39⁻ CD8⁺ T cell subsets. Gene set enrichment analysis (GSEA) revealed upregulation of genes associated with T cell exhaustion pathways in CD39⁺ cells, resulting in significantly higher exhaustion scores compared to CD39⁻ T cells. Moreover, CD39 expression levels were positively and significantly correlated with exhaustion scores. Multicolor flow cytometric analysis also demonstrated co-expression of exhaustion markers (PD-1 and TIGIT) on CD39⁺ CD8⁺ T cells. Stratification by PD-1 intensity revealed that a substantial subset of CD39⁺ CD8⁺ T cells fell into the PD-1hi category, consistent with a terminally exhausted phenotype. We next evaluated the clinical relevance of CD39⁺ T cells by analyzing their association with treatment response. Among patients who received standard 3+7 induction chemotherapy, higher frequencies of CD39⁺ T cells in both CD4⁺ and CD8⁺ compartments were significantly associated with lower complete response (CR) rates. Similarly, increased CD39⁺ T cell frequencies correlated with inferior event-free survival (EFS), highlighting the prognostic implications of CD39-defined immune dysregulation. Comprehensive clinical and mutational profiling revealed heterogeneous genetic backgrounds, but no specific mutation was strongly associated with CD39 expression. In conclusion, CD39 represents immunosuppressive and exhausted T cell subsets in AML, particularly among non-Treg CD4⁺ and CD8⁺ populations. Its enrichment in differentiated and functionally exhausted T cells, coupled with its association with adverse clinical outcomes, supports the role of CD39 as a functional biomarker of immune dysfunction and chemoresistance. The incorporation of CD39-based immune profiling may complement existing molecular risk stratification systems by capturing dynamic immun