T cells can specifically recognize FLT3-ITD-mutated AML cells and induce cell lysis by secreting IFN-γ, granzyme B, and perforin (16). In patients with FLT3-ITD mutations, the proportion of CD3+ T cells (1), including both CD4+ and CD8+ subsets, is significantly increased (35), which contrasts with the significantly reduced percentages of B cells, plasmablasts (1, 36), and NKT cells (37). However, despite the increase in T cells, the anti-leukemia immune response is impaired due to multiple immune escape mechanisms, such as the predominant expansion of T_regs among CD4+ T cells. Studies have shown that T_regs are significantly enriched in the BM and spleen of FLT3-ITD-mutant AML mice (16), suggesting that the increase in CD4+ T lymphocytes is primarily driven by the expansion of T_regs.

In addition, immune evasion is also achieved through the following mechanisms: upregulation of co-expressed immunosuppressive molecules on CD8+ T cells (38), and the elevated expression of immune checkpoint receptors like TIM-3 and LAG-3 (29). Although the proportion of CD8+ T cells is increased in these patients, they typically exhibit a TIGIT+PD-1+DNAM-1− phenotype (38). The co-expression of these immunosuppressive molecules is associated with poorer prognosis (39). However, although no significant differences in the expression of TIGIT and PD-1 were observed between the FLT3-ITD mutant and wild-type groups when analyzed in the overall T cell population in some samples (15), it is important to note that the immunosuppressive effects are primarily determined by the expression of immune checkpoint receptors on T cells that specifically recognize leukemia antigens (40). Additionally, following mutation, the FLT3 receptor remains aberrantly activated, and in combination with the effect of FLT3 ligand, this leads to elevated expression of TIM-3 in T cells (29). The autocrine or paracrine signaling pathways of TIM-3 promote leukemia cell proliferation and anti-apoptotic activity, while also suppressing the function of distant immune cells (29). TIM-3 expression is accompanied by galectin-9 secretion, which inhibits T cell activity. TIM-3 transcript levels correlate with CLIP levels (29), suggesting that immune evasion mechanisms are often co-activated. Moreover, In the context of FLT3-ITD mutation, the expression of the immune checkpoint LAG-3 is significantly increased in T cell subsets (27). LAG-3 impairs T cell receptor (TCR)-mediated signaling, thereby affecting the proliferation and function of T_effs cells. High LAG-3 expression is associated with shorter OS and disease-free survival in AML patients (27, 39) ( Figure 1 ).

Compared to wild-type AML patients, those with FLT3-ITD mutations exhibit a significant expansion of DCs, particularly common DC progenitors and precursor DCs. In mouse models, the effect of FLT3-ITD on DCs is allele dose-dependent; the more copies of the mutation present, the greater the expansion of DCs. This expansion promotes the proliferation of T_regs, a phenomenon that becomes especially pronounced in the BM as the mutation burden increases (36, 41). Concurrently, DCs undergo abnormal phenotypic changes. The frequency of XCR1/cDC1 double-negative cDCs is markedly elevated, and these cells display impaired antigen presentation capabilities (16). In FLT3-ITD mutant patients, CLIP on the surface of cDCs remains bound to and exposed on HLA molecules. CLIP is an invariant chain polypeptide essential for HLA class II antigen presentation and can also be cross-presented on HLA class I molecules. Persistent exposure to CLIP disrupts T cell activation and is associated with poorer prognosis (29). Additionally, compared to healthy mice, the cDC phenotype in FLT3-ITD mutant mice is skewed toward T-bet-expressing cDC2. Under the influence of specific cytokines, these cDC2 cells effectively polarize naive CD4+ T cells into Th17 cells, leading to increased production of IL-17A. This Th17 subpopulation has been linked to unfavorable prognosis in AML (16) ( Figure 2 ). Meanwhile, the immune evasion mechanisms are further complicated by alterations in TNF secretion by macrophages. The secretion of TNF by macrophages is decreased in patients with FLT3-ITD mutations (42), but TNF exerts a dual effect on tumor cells. Under normal conditions, TNF is involved in regulating T lymphocyte-mediated homeostasis and anti-tumor responses, thereby improving CR rates and extending event-free survival. However, studies have shown that lower levels of TNF may promote the death of tumor-infiltrating T cells, enhance tumor cell differentiation, and facilitate the migration of myeloid cells, thereby accelerating leukemia progression (1). Therefore, the role of TNF in AML is complex, and changes in its levels may have varying effects on disease progression.

The frequency of TIGIT+ M2 macrophages is elevated in AML patients with the FLT3-ITD mutation. The high infiltration and expression of TIGIT in M2 macrophages are significantly associated with poor prognosis in AML (14). In AML cells harboring both FLT3-ITD mutations and the KMT2A::MLLT3 fusion, the genes CCN3 and SOCS2 become activated (43). Activation of CCN3 recruits macrophages and promotes their differentiation into the M2 phenotype, downregulates the expression of CD36 and SRA1, and consequently reduces phagocytic function (44). Additionally, as a target gene of FoxO1, CCN3 activation enhances the functionality of T_regs (45). SOCS2 plays a multifaceted role by not only inhibiting the expression of pro-inflammatory cytokines and the development of T helper 2 (Th2) cells in DCs but also promoting the polarization of CD4+ T cells into inducible regulatory T cells (iTregs) (46) ( Figure 1 ). It maintains the stable expression of Foxp3 in iTregs by inhibiting the IL-4 signaling pathway, thereby sustaining the anti-inflammatory phenotype and cellular stability of iTregs (47). Moreover, excessive activation of SOCS2 leads to the inhibition of IL-8 secretion and upregulation of RANTES expression, both of which are associated with poor prognosis (48). These findings align with observations in AML, where elevated levels of SOCS2 correlate with reduced OS (43).

In addition, the proportion of NK cells in AML patients with FLT3-ITD mutations was significantly elevated, and the copy number, ITD length, and mutant allele frequency of FLT3-ITD mutations were positively correlated with the proportion of NK cells. However, patients with a high proportion of NK cells at the initial stage of AML tend to have a poor prognosis. This may be due to a reduced number of mature NK cells and their limited cytotoxic function (37). Furthermore, the expression of inhibitory receptors is generally increased across all NK cell subsets, although it exhibits heterogeneity (8). In a study by Cianga’s team analyzing BM samples from eight AML patients, the proportion of overmature NK cells in patients with FLT3 mutations was significantly increased (9). Conversely, in another study by the same team examining PB from 20 newly diagnosed AML patients, those with FLT3 mutations exhibited an extremely low proportion of NK cells and markedly abnormal expression levels of the inhibitory receptor CD159a (8). These discrepancies may be attributed to differences in patient cohorts, variations between the BM and PB environments, and the regulatory influence of the TME on NK cell development.

AML patients with FLT3-ITD mutations have increased numbers of mucosal-associated invariant T (MAIT) cells, which predominantly exhibit effector memory or terminally differentiated phenotypes (35), indicating high activation but also signs of aging and exhaustion, characterized by upregulated PD-1 and downregulated CD161 expression (49). Moreover, MAIT cell function is compromised, with reduced Th1-type cytokine production (IFN-γ and TNF-α) and increased secretion of Th17-type cytokines (IL-17A and IL-8), granzyme B, and perforin (49). We speculate that in AML patients, despite reduced Th1-type cytokine production, MAIT cells may exert anti-tumor effects primarily through granzyme B and perforin-mediated degranulation and cytokine activation, as indicated by the increased secretion of Th17-type cytokines and cytotoxic molecules. These findings suggest that MAIT cells have a dual role in AML, functioning both as anti-tumor agents and potentially as tumor promoters. However, the changes in PD-1 and CD161 expression and the secretion of immune effector molecules in the context of FLT3-ITD mutations require further investigation.

The abnormal cytoplasmic localization of NPM1c leads to the generation of novel neoepitopes, including AIQDLCLAV (AIQ) (5) and CLAVEEVSL (CLA) (53), which can be recognized by specific TCRs. The AIQ epitope, presented by HLA-A2, can bind to specific TCRs, and T cells engineered to express these TCRs effectively kill NPM1c+HLA-A2+ AML cells. AML patients with NPM1c+ who exhibit AIQ-specific CD8+ T cell responses have significantly longer survival (5). In contrast, the CLA epitope does not elicit a significant T cell response (53), suggesting that CLA may be suppressed by the TME in vivo. NPM1 mutation-specific CD8+ T cells can directly lyse leukemia cells harboring NPM1 mutations, whereas CD4+ T cells support CD8+ T cell function and induce HLA class II-mediated anti-tumor cytotoxic responses. These specific T cells predominantly express CD107a (54), indicating their activated state and involvement in cytotoxic activity ( Figure 3 ). Notably, specific T cell responses can still be detected after morphological CR in AML patients. This suggests that even when leukemic cells are substantially reduced, these T cells continue to eliminate minimal residual disease (MRD), thereby maintaining long-term CR and reducing relapse risk. A decrease in specific T cells is associated with disease relapse, and patients exhibiting these T cell responses have longer OS (54), further indicating a strong correlation with better prognosis. However, it remains to be investigated whether the frequency and intensity of NPM1 mutation-specific T cell responses vary based on patients’ molecular characteristics.

AML cells can still evade immune surveillance through multiple mechanisms. In particular, under the influence of neoantigenic epitopes produced by NPM1 mutations, leukemic cells predominantly express inhibitory B7 family ligands (such as PD-L1, PD-L2, B7.2, and VISTA) (51, 55). Elevated PD-L1 expression significantly enhances the immunosuppression of NPM1 mutation-specific CD8+ T cells (56). This high level of PD-L1 is strongly associated with poorer patient outcomes, in part due to the expansion of T_regs (57), especially in patients with concurrent FLT3-ITD mutations (51, 54). Although some studies suggest that NPM1 is essential for PD-L1 expression and that NPM1 mutations may slightly reduce PD-L1 expression, these differences are not statistically significant (58), and the underlying mechanisms remain to be elucidated. Furthermore, In patients with NPM1-mutated AML, TIM-3 transcript levels are also significantly reduced (58). TIM-3 plays a complex role in immune function, correlating with T cell exhaustion while also enhancing NK cell cytotoxicity. Some studies indicate that low TIM-3 expression may be associated with favorable prognosis in NPM1-mutated AML, whereas higher TIM-3 expression is linked to significantly lower CR and survival rates at one-year follow-up (58). These findings suggest that TIM-3 may serve as a potential indicator of poor prognosis. Notably, patients with a greater number of TIM-3+ NK cells exhibit better prognoses (59), emphasizing the dual nature of TIM-3 in AML. Overexpression of the pro-inflammatory mediator LTB4R is positively correlated with the expression of inhibitory immune checkpoint molecules such as PD-1 and TIM-3, while showing a negative correlation with immune effector cell populations (60). Although the mechanistic link remains unclear, this suggests that LTB4R may contribute to the immunosuppressive landscape in NPM1-mutated AML. In addition, NPM1 mutations have been shown to enhance the expression of CD47, a key ‘don’t eat me’ signal, which interacts with SIRPα on macrophages to inhibit phagocytosis. This mechanism further protects leukemic cells from immune clearance by the innate immune system (58). Although activating signals such as ULBP1 can stimulate NK and T cell responses through engagement with the NKG2D receptor, their immunostimulatory effects may be attenuated by the concurrent upregulation of immune checkpoint molecules, further enabling immune evasion in NPM1-mutated AML (51).

The antigen presentation process is also inhibited. In the absence of DNMT3A mutations, NPM1 mutations lead to the downregulation of the CIITA gene (51), thereby inhibiting the expression of CLIP protein and HLA molecules, which helps leukemia cells evade recognition by CD8+ T cells. However, AML cells partially retain HLA expression, and cells with low HLA expression do not exhibit higher NK cell lysis rates (58), suggesting that leukemia cells may balance NK and T cell attacks in immunoediting by regulating HLA levels. Furthermore, studies have found that the frequency of specific HLA-I alleles in patients with NPM1 mutations is significantly lower than that in healthy controls and NPM1 wild-type AML patients (54). This suggests that HLA alleles capable of effectively presenting NPM1 peptides may reduce the risk of developing NPM1-mutated AML. Even among patients carrying such alleles who develop the disease, specific immune responses may contribute to disease remission. DNMT3A mutations can weaken the effect of NPM1 mutations on HLA expression. Interestingly, in samples with high HLA-DR expression, NPM1 mutations are associated with higher CLIP levels, indicating a complex regulation of antigen presentation and immune responses (58).

In addition to modulating classical immune checkpoint pathways, NPM1 mutations suppress immune function through a range of noncanonical mechanisms. One such mechanism involves the regulation of small extracellular vesicle-mediated signaling. The NPM1c/CTCF/PABPC1 signaling axis controls the secretion of miR-19a-3p via small extracellular vesicles, which are subsequently internalized by CD8+ T cells. This process inhibits the expression of creatine transporters, prevents creatine uptake, reduces ATP production, and consequently impairs the immune function of CD8+ T cells (53). Moreover, NPM1 mutations significantly upregulate SPINK2 expression and downregulate ALCAM, both of which contribute to impaired T cell activation (61).

DNMT3A plays a pivotal role in shaping immune cell fate, particularly by maintaining the differentiation stability of CD4+ T cells and restricting both the formation of long-term memory CD8+ T cells and the pool of memory precursor effector cells (28, 67). In the context of AML, DNMT3A mutations have been associated with notable alterations in T cell infiltration and subset composition. In AML patients, DNMT3A mutations are associated with increased T cell infiltration; however, the distribution of T cell subsets is aberrant. Specifically, in patients with wild-type DNMT3A, T cells tend to undergo terminal differentiation, resulting in a reduced proportion of memory T cells (28). In contrast, patients with DNMT3A mutations exhibit a reduction in CD8+ T_n and CD4+ T_em in the BM, accompanied by an increase in CD4+ central memory T cells. Clinical studies have demonstrated that using donors with a higher proportion of CD8+ T_n for lymphocyte infusion can contribute to long-term remission in AML patients (28). Notably, similar to other mutations affecting DNA methylation regulators such as NPM1, IDH2, and CEBPA, DNMT3A mutations have also been shown to upregulate tumor-specific antigens, which in turn can activate antigen-specific clonal T cell responses (68). Conversely, lower ratios of CD8+ T_n and CD4+ T_em are associated with adverse genetic risks and poorer relapse-free survival and event-free survival (28). The specific impact of T_em cells on prognosis remains controversial. Some studies, such as those by Ling Xu and Adam J. Lamble, have found that an increased proportion of T_em cells is associated with enhanced T cell proliferation and higher CR rates. However, Maddalena Noviello’s team reported that the proportion of T_em cells also increases in relapsed AML patients (28). Moreover, DNMT3A mutations are linked to a higher risk of acute graft-versus-host disease following allogeneic hematopoietic stem cell transplantation, primarily by promoting CD4+ T cell polarization toward a Th1 phenotype and enhancing IFN-γ production (67). These findings collectively highlight the profound impact of DNMT3A mutations on T cell differentiation, function, and clinical outcomes in AML.

Beyond modulating adaptive immunity, DNMT3A mutations also disrupt innate immune signaling and promote an immunosuppressive TME. One such mechanism involves the hypomethylation-induced upregulation of miR-196b, which directly inhibits the Toll-like receptors (TLR) 7/8 signaling pathway, thereby weakening the immune response. In normal immune cells, TLR7 activates type I IFN and cytokines through the MyD88 pathway, indirectly activating Stat1 signaling to enhance the activity of Th1 cells and monocytes while promoting DC differentiation (69). Consequently, inhibition of TLR7/8 may lead to a diminished overall immune response. DNMT3A mutations are also frequently accompanied by increased infiltration of T_regs (70). Previous studies have shown that T_regs are highly adaptable to different tissue environments, and DNMT3A-dependent de novo DNA methylation facilitates this adaptability by establishing tissue-specific epigenetic memory, thereby refining and modulating their functions. However, in AML, DNMT3A mutations may impair methyltransferase activity, making it difficult for T_regs to adapt to various environments and thereby affecting their immune regulatory functions (71). Additionally, DNMT3A-dependent de novo DNA methylation is essential for silencing Foxp3 transcription. Mutations in DNMT3A may impair the effective silencing of Foxp3 transcription in T_regs, allowing them to continuously maintain their immunosuppressive functions, which may further promote immune escape in AML (72). In addition to T_reg-mediated suppression and impaired TLR signaling, DNMT3A-mutated AML cells also exhibit elevated levels of immunosuppressive cytokines such as IL-10 and TGF-β (62). Furthermore, the concurrent upregulation of immune checkpoint molecules including PD-L1, CLIP, and TIM-3 indicates a coordinated activation of multiple immune evasion pathways, collectively contributing to a highly suppressive TME and poor prognosis (58).

In addition to its effects on lymphoid immunity, DNMT3A mutation significantly alters the myeloid compartment. AML cells harboring DNMT3A mutations have been shown to possess an enhanced capacity to chemoattract monocytes, thereby modifying the TME to favor immune suppression. These AML cells inhibit the activity of the AP-1 binding site, leading to the downregulation of pro-inflammatory cytokines such as MIP-1α, MIP-1β, and IL-1β (62). The suppression of these key mediators impairs M1 macrophage polarization and diminishes their cytotoxic functions against tumor cells. In vivo studies using murine models have demonstrated a marked increase in the proportion of M2-polarized tumor-associated macrophages (TAMs) in the presence of DNMT3A mutations. These M2 macrophages express high levels of CD163 and CD206 and secrete chemokines such as CCL17, CCL22, and CCL24, which recruit Th2 cells to the leukemia microenvironment (62). The accumulation of M2 TAMs and Th2 cells creates an anti-inflammatory milieu that correlates strongly with reduced patient survival. Furthermore, DNMT3A-mutated AML cells exhibit resistance to macrophage-mediated killing and can differentiate into monocyte-like cells with immunosuppressive properties, further contributing to the inhibition of effective anti-leukemic T cell responses (62). These findings underscore a critical role for DNMT3A mutations in reprogramming the myeloid landscape, skewing macrophage polarization toward an immunosuppressive phenotype, and establishing a tumor-permissive microenvironment.

TP53 mutations in AML lead to substantial alterations in the function and composition of T cells. Despite the increased infiltration of T cells, these cells exhibit signs of exhaustion and impaired functionality (71). Specifically, the expression of activation markers such as CD25, HLA-DR, and CD127 is low (74), while the immune checkpoint receptor CTLA-4 is upregulated (75). Moreover, the expression of cytotoxic molecules such as perforin and granzyme B is diminished (76), and the secretion of Th1 cytokines is significantly reduced (77). The T cell subsets are also significantly altered. Unsupervised clustering analysis revealed that CD8+ T cells predominantly exhibit an ICOS+/4-1BB+/PD-1+ phenotype (78), suggesting a dysfunctional, exhausted state. In contrast, Th cells predominantly express ICOS (6), which, despite being a costimulatory molecule, contributes to immune evasion in this context. The increase in PD-1+ cytotoxic T lymphocytes and PD-L1+ BM blasts further supports this hypothesis, as PD-1 signaling has been shown to inhibit T cell function and induce exhaustion (79). The upregulation of PD-L1 is closely associated with the downregulation of miR-34a and the overexpression of the MYC gene. Under normal conditions, wild-type p53 induces the transcription of miR-34a, which targets MYC mRNA and promotes its degradation, thereby negatively regulating MYC expression. However, in TP53-mutated AML, miR-34a expression is significantly reduced, leading to the upregulation of MYC and the induction of PD-L1 expression. The downregulation of miR-34a weakens its binding to the 3′ untranslated region of PD-L1 mRNA, reducing the inhibition of PD-L1 expression and thereby promoting T cell exhaustion (6, 78) ( Figure 4 ). Transcriptional analysis revealed that, compared to healthy controls, cytotoxic T lymphocytes from TP53-mutated AML patients exhibited upregulation of inhibitory molecules (such as CD244, CD160, LILRB1, CD300A, and PVRIG) and downregulation of stimulatory molecules (such as CD40LG, CD28, TNFSF8, TMIGD2, and TNFRSF25). These alterations were not significant in other AML molecular subtypes (79).

In contrast, T_regs demonstrated metabolic adaptations and proliferative advantages. Although OX40+ T_regs are markedly reduced in the BM, the highly immunosuppressive ICOShigh/PD-1neg T_regs are significantly expanded (6). These T_regs exhibit enhanced proliferative capacity and are implicated in the suppression of anti-tumor immunity, which has been identified as an independent predictor of poor OS (78). The IL-2/STAT5 signaling axis further promotes T_reg differentiation and stabilization (74, 80), with elevated FOXP3 expression (74) and increased secretion of immunosuppressive cytokines such as TGF-β and IL-10 (76). Together, these findings illustrate a significant alteration in T cell functionality, with exhaustion of T_effs and expansion of suppressive T_regs, contributing to the immune escape observed in TP53-mutated AML.

TP53 mutations in AML significantly affect innate immune and leads to the formation of a pro-inflammatory microenvironment. One important mechanism is the downregulation of HLA molecules, which impairs antigen presentation, thus preventing the immune system from effectively recognizing and eliminating leukemia cells (74, 76). In addition, despite promoting significant infiltration of TAMs (80), the overexpression of CD47 on leukemia stem cells interacts with the SIRPα receptor on TAMs, inhibiting their phagocytic activity (81). Additionally, TP53 mutations lead to the upregulation of JAK/STAT, PI3K-Akt, and NF-κB signaling pathways, which are associated with increased production of pro-inflammatory cytokines such as CXCL1, CXCL2, CXCL8/IL-8, and IFN-induced products like CCL2, IL33, and IL6 (75). This results in the formation of a pro-inflammatory microenvironment, with IFN-γ playing a dominant role, which has been linked to poor responses to induction chemotherapy. Moreover, TP53 mutations inhibit IRF3 transcriptional activity and affect IFN expression through two mechanisms (1): by binding to TBK1, preventing the formation of the STING-TBK1-IRF3 complex (82), and (2) by inducing the overexpression of PLK4, which further inhibits the activation of the cGAS-STING-TBK1-IRF3 pathway (83).