Acute myeloid leukemia (AML) is a clonal disorder characterized by immature blasts and arrested differentiation of malignant myeloid blasts in the bone marrow. Advances in understanding AML's genetic underpinnings have improved our understanding of its biology, prognosis, and treatment, leading to FDA approvals of targeted therapies. Venetoclax, a BCL2 inhibitor, has been transformative, particularly in older patients when combined with azacitidine or low-dose cytarabine. While effective in many, a significant portion of patients exhibit primary resistance or early relapse. The mechanisms of venetoclax resistance are not fully understood, but monocytic subclones are implicated. The interplay between the bone marrow immune microenvironment and AML pathogenesis is also incompletely understood. Dysregulated inflammatory pathways are linked to leukemogenesis and the maintenance of leukemic blasts. While some studies suggest AML-intrinsic repression of inflammation contributes to cell survival, others indicate that pro-inflammatory cytokines in the microenvironment can induce leukemogenesis and impact hematopoietic cell proliferation. Type 1 and 2 interferons are key players in the inflammatory response to cancers. Interferon-alpha (IFNα) is FDA-approved for certain leukemias and myeloproliferative neoplasms, and IFNγ treatment has been explored in post-transplant AML patients. However, IFNγ has also been shown to promote cancer growth, suppress T cell activity, and drive an exhausted phenotype, highlighting its complex and dichotomous role in cancer. This study utilizes bulk and single-cell approaches to investigate the relationship between AML blasts and their immune microenvironment, focusing on IFNγ signaling and its association with venetoclax resistance.

The literature review section comprehensively examines the existing knowledge on interferon gamma (IFNy) signaling in acute myeloid leukemia (AML). It covers studies highlighting IFNy's dual role in immune regulation and cancer progression, referencing its diverse functions in immune regulation, inflammation, and tumor surveillance, alongside studies showing its potential to both suppress and promote tumor growth depending on the context. The review incorporates research on type 1 interferons (IFNα, IFNβ, IFNω) and IFNγ, discussing their roles in the inflammatory response to cancer and their clinical applications in various hematological malignancies. It also covers the use of IFNα in myeloproliferative neoplasms and chronic myeloid leukemia, as well as investigations into IFNγ's potential in post-transplant AML patients. The review incorporates conflicting evidence about IFNy's impact on cancer, noting both its antitumor and pro-tumorigenic effects, such as its contribution to immune evasion, exhausted phenotype, and immune checkpoint receptor induction. The literature lays the groundwork for the study by emphasizing the existing knowledge gaps regarding the complex interplay of IFNy signaling within the AML microenvironment and its potential implications for treatment.

This study employed a multi-faceted approach combining bulk and single-cell RNA sequencing (scRNA-seq) data from newly diagnosed AML patients. Bulk RNA-seq data from three independent datasets (TCGA, MDACC, and BEAT-AML) were used to assess IFNγ signaling using single-sample gene set enrichment analysis (ssGSEA). Correlation analysis was performed to identify relationships between IFNγ signaling scores and clinical characteristics, including blast percentage, FAB classification, cytogenetic groups, and HLA expression. Immune deconvolution with CIBERSORTx was used to correlate cellular composition with IFNγ signaling. scRNA-seq was conducted on bone marrow aspirates from 20 AML patients. Data processing involved quality control, normalization, and batch effect correction using Seurat and Harmony. InferCNV was used to infer copy number variations. Canonical marker genes and flow cytometry were used to validate AML cell identity and differentiation states. SCENIC was utilized to infer gene regulatory networks and regulon activities. CellChat and MultiNicheNet were used to analyze cell-cell interactions. Spectral flow cytometry was employed to validate HLA expression on AML blasts and multiplex immunofluorescence was used to analyze the spatial organization of the bone marrow microenvironment. For functional studies, AML blasts from primary patient samples and AML cell lines were used in co-culture assays and IFNγ stimulation assays to assess IFNγ production and IFITM3 expression. DepMap CRISPR knockout data were used to assess IFITM3 dependency in AML cell lines. LASSO regression was used for feature selection to create a parsimonious IFNγ signature. Statistical analyses including Pearson correlation, Kruskal-Wallis, Wilcoxon, log-rank, and Cox regression were employed to assess the significance of findings. The study was conducted in compliance with the Declaration of Helsinki and obtained informed consent from all participants.

The study revealed a high IFNγ signaling activity in monocytic and del7/7q AML subtypes, identified through both bulk RNA-seq (using ssGSEA) and scRNA-seq. The highest IFNγ signaling scores were observed in monocytic AML cells, correlating strongly with the presence of IFNγ-producing T and NK cells. This high IFNy signaling in monocytic AML was validated in two independent scRNA-seq cohorts. Bulk analysis revealed a strong positive correlation between IFNγ signaling and HLA class I and II expression, and associations with T cell dysfunction, exhaustion, and senescence. Single-cell analysis demonstrated that IFNγ signaling was predominantly within AML cells, being highest in monocyte-like AML cell states. Analysis using SCENIC identified different regulon activities of interferon regulatory factors (IRFs) across cytogenetic subgroups. Cellular communication analyses (CellChat and MultiNicheNet) highlighted prominent T cell-AML interactions, with IFNγ from CD8 T cells and NK cells being predicted as a top interaction with AML cells in diploid monocytic AML. Co-culture assays showed that AML cells, when co-cultured with T cells, led to a significant increase in IFNγ secretion. Immunofluorescence further confirmed that HLA-E+ CD34+ AML cells were in closer proximity to CD3+ T cells. The study found a strong positive correlation between IFNγ signaling and venetoclax resistance, confirmed in an independent cohort. Treatment of primary AML cells with IFNγ increased venetoclax resistance. IFITM3 was identified as a downstream target of IFNγ, with its expression positively correlating with IFNγ signaling and independently predicting worse overall survival in AML patients. Knockout of IFITM3 decreased the fitness of all AML cell lines tested. Finally, a parsimonious 47-gene IFNγ signature derived using LASSO regression showed robust prognostic value, independently predicting worse survival in AML patients.

This study provides a comprehensive characterization of the complex interplay between IFNγ signaling and the immune microenvironment in AML, offering significant insights into disease progression and therapeutic resistance. The findings highlight the crucial role of IFNγ in shaping a distinct immunosuppressive microenvironment, particularly in monocytic AML. The strong correlation between IFNγ signaling and venetoclax resistance suggests that targeting IFNγ could be a potential strategy to overcome this resistance. The identification of IFITM3 as a downstream target of IFNγ and its independent prognostic value adds another layer of understanding to AML biology. The study's findings support the hypothesis that IFNγ contributes to immune evasion, possibly through the upregulation of HLA-E, suggesting potential therapeutic strategies focused on modulating the HLA-E-NKG2A axis. This study underscores the importance of considering the complex interplay between AML cells and the immune microenvironment in developing effective therapeutic strategies.

This study's integration of bulk and single-cell RNA sequencing data revealed high IFNγ signaling in monocytic and del7/7q AML, correlating with venetoclax resistance and poorer prognosis. IFITM3 emerged as a key downstream target and potential therapeutic vulnerability. A parsimonious IFNγ signature proved a strong independent prognostic indicator. These findings suggest that targeting IFNγ signaling, potentially through modulating IFITM3 or the HLA-E/NKG2A axis, could be a valuable therapeutic strategy in AML.

While this study provides extensive insights, several limitations should be noted. The *in vitro* models may not fully capture the intricate dynamics of the bone marrow microenvironment. Further research is needed to directly validate the role of IFITM3 in mediating venetoclax resistance and the impact of IFNγ in an immunocompetent model. While IFNγ's prominence is highlighted, other inflammatory pathways might also play a role, necessitating further investigation. The study’s focus on IFNγ does not exclude the potential influence of other inflammatory pathways, necessitating future research to explore these additional contributing factors.