Dynamics of leukemic blast and immune cell populations in Acute Myeloid Leukemia

Abstract Although patients with acute myeloid leukemia (AML) initially respond to standard chemotherapy, the relapse rate is high, and many subsequently develop refractory disease. This is largely due to the clonal heterogeneity and molecular diversity of the disease, such that drug resistant subclones easily develop. Thus, response to targeted therapies can be transient while the efficacy of immunotherapy remains limited. To expand treatment options, a comprehensive understanding of the clonal landscape and immune microenvironment of AML, as well as the interactions between immune and leukemic cells is essential. Previous studies utilizing single-cell RNA sequencing (scRNA-seq) have provided valuable insights, but they predominantly focused on limited, well-controlled or newly diagnosed patient cohorts. Here, we instead characterized the cellular composition, immune marker landscape, and dynamics of leukemic cells, as well as their interactions with immune cell populations at single cell resolution in a cohort of AML bone marrow (BM, n=71) and peripheral blood (n=1) samples from patients with mixed treatment histories, sampled at diagnosis (Dg, n=22), remission (Rm, n=2), and relapse/refractory (RR, n=48) stages. Among these were 26 Dg-RR, 6 Dg-Rm-RR, and 4 RR-RR longitudinal sample sets. All samples were obtained through the Finnish Hematology Research Biobank, following protocols approved by an ethical committee of the Helsinki University Hospital and in accordance with the Declaration of Helsinki. ScRNA-seq libraries were prepared using 10X Genomics kits and Chromium instrument, followed by sequencing on the Illumina NovaSeq 6000 instrument. Data analysis was performed using Seurat. Cell types were identified by mapping our dataset to the bonemarrowmap reference dataset (Zeng, A. et al., 2025), and leukemia-specific markers were used to identify blast cells. A publicly available healthy BM dataset was retrieved for comparison (Oetjen, K.A. et al., 2018). As expected, cellular composition was highly variable among the samples. Compared to Rm and healthy BM, Dg and RR AML samples had significantly higher proportions of progenitor cells and lower proportions of immune cells, suggestive of treatment-related changes. We had 21 and 7 RR samples from patients treated previously with cytarabine and venetoclax, respectively. Phenotypic changes associated with induction chemotherapy included loss or reduced density of HSC/MPPs or complete change in phenotype. An example of the latter included a patient who initially had a high density of HSC/MPPs at diagnosis, followed by a loss of HSCs/MPPs, a slight increase in GMPs and CLPs, and differentiated monocytes at remission. Upon relapse, the patient showed a high density of GMPs/CLPs, with a loss of monocytes and unrecovered HSCs. In longitudinal RR samples from cytarabine-treated patients, differential gene expression and pathway analyses of blast cells revealed increased mitochondrial function and energy metabolism, including oxidative phosphorylation and ATP synthesis, along with suppressed inflammatory response pathways compared to diagnosis. Blasts from venetoclax-treated patients showed increased erythroid differentiation at relapse. We also compared markers of immune cell exhaustion and activation in healthy and AML datasets as well as during disease progression. In AML dataset, we detected PD-1 expression in small immune cell clusters, including expression in approximately 20% of proliferating T cells and about 10% in CD4+ regulatory T cells. We observed that VISTA, an inhibitory checkpoint protein, was expressed in AML Dg and RR samples but absent in Rm and healthy BM. VISTA restrains T cell-mediated anti-tumor immunity and could be explored as a potential immunotherapeutic target for AML. Cell-cell communication analysis between blasts and immune cell population identified multiple interactions between leukemic stem cells ligands and inhibitory receptors in NK cells including NECTIN2-CD226, LGALS9 (galectin9) - HAVCR2 (TIM3), HLA-E - CD94/NKG2A, CD94/NKG2B, CD94/NKG2C, CLEC2B-KLRB1. These findings underscore the dynamic changes in AML cell populations in response to treatment and during disease progression, implicate potential relapse mechanisms and the interactions used by leukemic stem cells to inhibit immune cell function and promote immune tolerance, and highlight novel treatment strategies.