Search Results (2,638)

The bone marrow tumor microenvironment (TME) is critical for acute myeloid leukemia (AML) progression, immune evasion, and treatment resistance. SRC, a non-receptor tyrosine kinase involved in multiple oncogenic pathways, has not been systematically characterized in AML in relation to prognosis and immune regulation. We integrated bulk transcriptomic and single-cell RNA-sequencing datasets from TCGA, BeatAML, and GEO. Immune-related targets were identified using xCell-based immune scoring and weighted gene co-expression network analysis (WGCNA), followed by protein–protein interaction analysis and multi-algorithm machine-learning screening. We then evaluated SRC expression patterns, prognostic associations, immune microenvironment features, predicted drug sensitivity, single-cell differentiation dynamics, intercellular communication, and in silico virtual knockout perturbation (scTenifoldKnk). SRC emerged as the most robust hub gene after integration of WGCNA, PPI analysis, machine-learning feature selection, and survival screening. SRC was significantly upregulated in AML compared with normal controls and was independently associated with poor overall survival (HR = 1.231, p = 0.037). High SRC expression was linked to adverse ELN/FAB features, increased immune checkpoint expression, enrichment of inflammatory and immunoregulatory pathways, and a higher proportion of primitive leukemia-associated cell states. Single-cell analyses further suggested that SRC was enriched in CD34⁺ progenitor compartments, associated with altered cell–cell communication, and accompanied by distinct mutation and pathway profiles. Drug-response prediction and in silico network perturbation analysis further supported the potential biological and translational relevance of SRC-centered programs. SRC is a prognostically relevant and immune-associated hub linked to AML microenvironment remodeling, and may serve as a candidate biomarker and potential therapeutic target that warrants further experimental validation. Full article

Background: Chimerism analysis is a key tool for monitoring donor-cell engraftment and the risk of relapse and graft-versus-host disease (GVHD) following allogeneic hematopoietic cell transplantation (allo-HCT). The advantage of lineage-specific chimerism assessment, and its dynamics in patients receiving post-transplant cyclophosphamide (PTCy)-based GVHD prophylaxis, remains unclear. Objective: This review summarizes the current state of the art on chimerism analysis in patients with myeloid neoplasms undergoing allo-HCT with PTCy, with emphasis on lineage-specific testing and modern methodologies. Methods: A structured literature review was conducted to assess chimerism dynamics in whole blood (WB), bone marrow, and peripheral blood (PB) subpopulations, including T-cells, CD34⁺, myeloid, B, and NK (natural killer) cells, and their association with clinical outcomes following PTCy. Results: Lineage-specific PB chimerism, particularly in T-cells, myeloid lineage and CD34⁺ cells, is more sensitive than WB chimerism for predicting relapse. Declining donor myeloid chimerism or persistent myeloid mixed donor chimerism (MDC) may precede hematologic relapse and provide an early signal of graft instability or ineffective graft-versus-leukemia activity. T-cell MDC has been associated with an increased risk of relapse and a lower risk of GVHD, although persistent T-cell MDC in some patients may instead indicate immune tolerance. Declining CD34⁺ donor chimerism correlates with a higher risk of relapse and inferior survival outcomes and may therefore complement measurable residual disease testing. Data regarding B-cell and NK-cell chimerism remain inconsistent, likely influenced by delayed immune reconstitution. Compared to anti-thymocyte globulin, PTCy may promote higher donor T-cell chimerism, though findings across studies are variable. Next-generation sequencing (NGS) enables more sensitive detection of microchimerism and relapse prediction. Conclusions: Chimerism analysis, particularly when lineage-specific and NGS-based, offers valuable prognostic insight in allo-HCT with PTCy. Further prospective studies are needed to standardize testing and guide personalized post-HCT strategies. Full article

Introduction: Chimeric antigen receptor (CAR) T-cell therapies have revolutionized treatment for relapsed/refractory hematologic malignancies, targeting CD19 in B-cell neoplasms and BCMA in multiple myeloma, with response rates exceeding 80%. However, long-term risks, including therapy-related myeloid neoplasms, such as myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML), are emerging 6–24 months post infusion, potentially linked to lymphodepleting chemotherapy, clonal hematopoiesis expansion, and inflammatory milieus. This FAERS pharmacovigilance analysis quantified MDS/AML reporting across seven FDA-approved CAR-T products to detect antigen-specific signals unattainable in pivotal trials with limited follow-up. Methods: Adverse event reports from FAERS (1 January 2013–10 February 2025) were queried for tisagenlecleucel, axicabtagene ciloleucel, brexucabtagene autoleucel, lisocabtagene maraleucel, obecabtagene autoleucel, idecabtagene vicleucel, and ciltacabtagene autoleucel, focusing on MedDRA terms for MDS/AML. Duplicates and ambiguous cases were excluded. Disproportionality was assessed using reporting odds ratios (RORs; lower 95% CI >1 signaling significance), comparing CAR-T-event pairs to database background, with subgroup analyses by antigen target. Results: Among 14,093,557 reports, CAR-T products linked to 303 MDS (brexucabtagene autoleucel ROR 97.93 [72.18–132.87], n = 44; axicabtagene ciloleucel ROR 58.70 [50.34–68.44], n = 172) and 129 AML cases (axicabtagene ciloleucel ROR 22.89 [18.23–28.73], n = 76). Signals were consistent across CD19- and BCMA-directed agents, absent only for recently approved obecabtagene autoleucel. Conclusions: CAR-T therapies exhibit disproportionate MDS/AML reporting in FAERS, supporting class-wide late hematologic toxicity in pretreated patients with clonal hematopoiesis. Enhanced surveillance, baseline profiling, and marrow evaluation for cytopenias are warranted, balancing curative benefits. Full article

Objective: To evaluate whether documented bacterial infection or neutropenic fever is associated with delayed hematopoietic recovery in patients with acute leukemia undergoing induction or consolidation chemotherapy. Methods: We conducted a retrospective observational study of 171 adult patients with acute myeloid leukemia (AML) or acute lymphoblastic leukemia (ALL) treated between 2022 and 2025. Patients were divided into three groups: (1) microbiologically documented infection (n = 73); (2) neutropenic fever without pathogen identification (n = 73); and (3) no fever or infection (n = 25). Hematopoietic recovery was assessed by time to neutrophil recovery (>0.5 × 10⁹/L) and time to reticulocyte production index (RPI) > 1.0. Statistical comparisons were performed using ANOVA or Kruskal–Wallis tests as appropriate. Results: The mean time to neutrophil recovery was 32.0 days (95% CI: 30.0–34.0) in Group 1, 28.4 days (95% CI: 27.3–29.6) in Group 2, and 15.2 days (95% CI: 14.3–16.2) in Group 3 (p = 0.0039). The mean time to RPI > 1.0 was 36.3 days (95% CI: 34.2–38.4), 33.0 days (95% CI: 31.8–34.2), and 19.4 days (95% CI: 18.5–20.3), respectively (p = 0.0018). Differences between Groups 1 and 2 were not statistically significant. Conclusions: Infection and neutropenic fever are associated with significantly prolonged hematopoietic recovery following chemotherapy for acute leukemia. Delayed regeneration may increase the risk of complications and negatively affect treatment outcomes. Full article

Background: Acute myeloid leukemia (AML) with monocytic differentiation poses significant clinical challenges, including high relapse rates and chemotherapy resistance. Current morphological assessment is limited by inter-observer variability, low sensitivity, and inefficiency, especially for detecting low-level residual disease. This creates an urgent need for automated, objective tools to improve diagnostic consistency and monitoring. Artificial intelligence, particularly deep learning, offers potential for extracting high-dimensional cytomorphological features to address these gaps. Methods: A retrospective cohort of 184 bone marrow smear slides from patients with monocytic leukemia was used. The core biomarker was the immature monocyte percentage (IMMP), defined as monoblasts plus promonocytes among nucleated cells, with a 2.0% clinical cutoff. An EfficientNet-based convolutional neural network was developed via transfer learning and trained to classify four cell types: monoblasts, promonocytes, monocytes, and other cells. Results: The model achieved robust cell-level classification, with F1 scores of 0.82 for monoblasts and 0.34 for promonocytes. At the slide level, using an optimized IMMP threshold of 0.045, it accurately assessed persistent leukemic cell burden with 78.9% Accuracy, 81.1% Recall, and 76.9% Specificity. Model-predicted IMMP values showed strong correlation with expert-derived values (Pearson r = 0.827), demonstrating reliable quantitative agreement. Conclusions: This deep learning model provides an automated, objective tool for quantifying immature monocytes, addressing key limitations in morphological assessment of monocytic AML. The IMMP metric shows promise for monitoring treatment response, predicting relapse, and potentially identifying patients at risk of venetoclax-based therapy resistance. While promising, prospective multicenter validation is needed to translate these findings into routine clinical practice. Full article

Acute myeloid leukemia (AML) remains a highly morbid malignancy in which outcomes are constrained not only by disease refractoriness and relapse, but also by therapy-related toxicity—particularly infections, mucosal injury, and delayed hematopoietic reconstitution. The gut microbiota has emerged as a potentially modifiable layer of host vulnerability and resilience during AML treatment. Microbiome disruption is detectable already at diagnosis, even in antibiotic-naïve patients, and is often characterized by reduced community diversity, depletion of anaerobic taxa linked to short-chain fatty acids (SCFAs) production, and enrichment of pathobiont-associated profiles. During induction, cytotoxic therapy and antimicrobials precipitates diversity loss, domination events, and persistent shifts beyond discharge. Clinically, the most consistent translational signal is the association between baseline or early-treatment microbiome features and infectious outcomes, while emerging data suggest that diagnosis-time microbiome structure may also relate to hematologic recovery kinetics. Mechanistic models converge on pathways linking barrier integrity, microbial metabolites (notably butyrate and other SCFAs), immune calibration, and inflammatory translocation of microbial products. These insights support hypotheses: antimicrobial stewardship may preserve microbiome function; ecosystem repair strategies such as autologous fecal microbiota transfer (A-FMT) are feasible and can restore community structure; and metabolite or nutritional interventions merit evaluation in immunocompromised hosts. Regimen-specific microbiome effects and microbiome–drug interactions suggest that treatment choice could have downstream microbiome-mediated consequences. We synthesize evidence, outline interventional concepts, and define methodological priorities for next-generation trials assessing causality and clinical benefit. Progress will require longitudinal sampling, multi-omic integration (metabolomics, resistomics, and barrier/inflammatory biomarkers), and interventional designs linking microbiome dynamics to clinically meaningful outcomes. Full article

Background/Objectives: Immunocompromised children undergoing chemotherapy or allogeneic hematopoietic stem cell transplantation (HSCT) for hematologic disorders face a high risk of serious, life-threatening infections caused by multidrug-resistant (MDR) bacteria. Cefiderocol is a novel siderophore cephalosporin, indicated for use in adult patients with MDR Gram-negative infections. Clinical data in immunocompromised children are limited. To report a multicenter real-life experience from the Infection Working Group of the Italian Pediatric Hematology and Oncology Association (IWG-AIEOP) on the use of cefiderocol in treating pediatric onco-hematologic patients with severe, high-risk infections. Methods: Multicenter retrospective collection of infectious episodes treated with cefiderocol, from January 2021 to December 2024, in patients 18 years or younger, after treatment for malignancies or undergoing HSCT in the AIEOP network, part of a prospective, observational study on the etiology and outcome of febrile episodes among 24 AIEOP centers (code NCT06419426). Results: Fifteen episodes of MDR, life-threatening Gram-negative infections treated with cefiderocol in 13 pediatric onco-hematologic patients were collected. There were eight males and five females, mainly affected by acute leukemia (six lymphoblastic and four myeloid, three other hematologic malignancies). The median age was 11.1 years (range 1–17.4 years), and the median weight was 37.8 kg (range 8–65). Bloodstream infection occurred in 10 of 15 episodes. Pseudomonas aeruginosa, Klebsiella pneumoniae, and Stenotrophomonas maltophilia were isolated in 11, 3, and 1 episodes, respectively. Notably, 11 of 15 isolated pathogens carried a metallo-beta-lactamase (MBL) gene (Verona integron-encoded, VIM, n = 10; New Delhi, NDM, n = 1). All patients achieved infection resolution and were alive and infection-free 90 days after infection onset. Conclusions: Cefiderocol was well tolerated and showed encouraging, favorable clinical outcomes, without serious adverse effects. Full article

Breakthrough invasive fungal infections (bIFIs) are a challenging serious complication in high-risk hematologic patients and allogeneic hematopoietic stem cell transplantation recipients that may negatively impact their outcome. Despite advances in antifungal prophylaxis, diagnostics, and supportive care, bIFI occurrence reflects a complex interaction between host immunosuppression, emergence of resistant pathogens and pharmacological variables, including subtherapeutic drug exposure. Candida spp. have shifted towards non-albicans yeasts, whereas breakthrough mold infections more frequently involve non-fumigatus Aspergillus, Mucorales, Fusarium spp., and Scedosporium/Lomentospora spp. Early clinical recognition, rapid therapy escalation, aggressive diagnostic investigation, a switch to liposomal amphotericin B-based regimens in patients on azole prophylaxis, and therapeutic drug monitoring are essential to improve outcomes. Reducing the growing global burden of bIFIs will also require improved access to high-quality diagnostics and strengthened educational and stewardship efforts that prioritize antifungal resistance as an urgent health concern. Full article

Background/Objectives: Myelodysplastic syndromes (MDS) are associated with a significant risk of progression to acute myeloid leukemia (AML), affecting approximately 30% of patients. In high-risk MDS, leukemic transformation may occur within a short time frame, highlighting the need for early and reliable biomarkers of disease progression. Increasing evidence suggests that immune dysregulation and cytotoxic T-cell dysfunction contribute to disease evolution. This study aimed to evaluate PD-1 and CD57 expressions on CD8⁺ T cells and to investigate the CD8⁺PD-1⁺/CD4⁺PD-1⁺ ratio (PERLS) as a potential immunological marker predictive of leukemic transformation. Methods: Thirty-one patients with MDS were prospectively followed over a 12-month period. At baseline, patients underwent routine clinical and laboratory evaluation, including multiparameter flow cytometric assessment of bone marrow blasts. An extended immunophenotypic analysis of bone marrow samples was performed at study entry to assess PD-1 and CD57 expression on CD8⁺ T cells. Cytogenetic and molecular analyses were conducted when clinical findings suggested disease progression. Patients who developed signs of progression were re-evaluated approximately one month later, during the progression phase, to assess dynamic immunological changes. Results: Of the thirty-one patients included, eighteen progressed to AML, whereas thirteen remained clinically stable. Patients who progressed demonstrated a significant increase in PD-1 and CD57 expression on CD8⁺ T cells compared with stable patients. Moreover, a markedly higher CD8⁺PD-1⁺/CD4⁺PD-1⁺ (PERLS) ratio was observed in patients who subsequently developed AML, particularly during the progression phase. Conclusions: Dynamic immunophenotypic monitoring reveals that increased PD-1 on CD8⁺ T cells and an elevated PERLS ratio are associated with imminent leukemic transformation in MDS. These findings support the incorporation of immune-based biomarkers, particularly the CD8⁺PD-1⁺/CD4⁺PD-1⁺ ratio, into routine risk assessment to enable earlier identification of disease progression and timely therapeutic intervention. Full article

Background: Graft-versus-host disease is the most common complication after allogeneic hematopoietic stem cell transplantation. While ocular graft-versus-host disease typically manifests as dry eye syndrome and anterior segment involvement, posterior segment complications are rare. Previously reported posterior segment complications in graft-versus-host disease have been limited to acute presentations with significant functional visual impairment. Methods: A 41-year-old man developed progressive retinal nerve fiber layer and ganglion cell layer loss four years after allogeneic hematopoietic stem cell transplantation for acute myeloid leukemia. The patient had established chronic graft-versus-host disease with cutaneous involvement and ocular surface disease. Results: Despite preserved visual acuity and visual fields, and only subtle functional involvement on visual evoked potentials, optical coherence tomography revealed significant reduction in retinal nerve fiber layer thickness and ganglion cell layer. Magnetic resonance imaging showed no optic nerve or brain abnormalities. Conclusions: This case describes an uncommon presentation of chronic, subclinical posterior segment involvement in chronic GVHD and suggests that optical coherence tomography may detect progressive structural retinal changes in the absence of clinically evident visual impairment, supporting its potential role in longitudinal monitoring. Full article

Philadelphia chromosome-positive acute lymphoblastic leukemia (Ph+ ALL) remains a high-risk entity despite advances in tyrosine kinase inhibitors (TKIs), immunotherapy, and cellular therapies. Relapse driven by clonal evolution, central nervous system (CNS) sanctuary disease, and TKI resistance, particularly T315I mutations, continues to limit durable disease control. Asciminib, a first-in-class allosteric BCR::ABL1 (STAMP) inhibitor, has demonstrated efficacy and favorable tolerability in chronic myeloid leukemia, but its optimal role in Ph+ ALL remains to be defined. We report a three-patient case series of Ph+ acute leukemia treated with asciminib across diverse high-risk clinical settings, including multiply relapsed disease, CNS involvement, T315I-mutated leukemia, post-CAR-T-cell relapses, and transplant bridging. Clinical outcomes are contextualized through a comprehensive review of emerging clinical trial data, real-world cohorts, and mechanistic studies evaluating asciminib in Ph+ ALL. Across all cases, asciminib was incorporated as part of combination or consolidation strategies rather than as monotherapy in active disease. Asciminib contributed to molecular disease control, CNS leukemia clearance, and successful bridging to allogeneic transplantation or cellular therapy, with acceptable tolerability and no major vascular toxicity. Integration of published evidence demonstrates that asciminib exhibits consistent biological activity in Ph+ ALL, with improved durability when used in rational combinations, particularly with immunotherapy or ATP-competitive TKIs. Preclinical data further support asciminib’s compatibility with antibody-based and cellular therapies through preservation of immune effector function. Asciminib represents a versatile but context-dependent therapeutic option in Ph+ ALL. Its greatest clinical value appears to lie in rational combination regimens, maintenance strategies, and bridging to definitive therapies rather than single-agent salvage. Emerging structural biomarkers and ongoing clinical trials are expected to further refine patient selection, sequencing, and optimal integration of asciminib, particularly in CNS-involved disease and post-CAR-T cell relapse. Full article

Background/Objectives: In recent years there has been a consistent development of clinical studies surrounding the incorporation of the B-cell lymphoma 2 (BCL-2) inhibitor venetoclax (VEN) into the treatment of acute myeloid leukemia (AML) Methods: A search of the literature showed a tremendous development of experimental and clinical studies evaluating the impact of VEN-based regimens in the treatment of AML patients. This review comprehensively analyzes the available scientific evidence—including prospective clinical trials, retrospective cohorts, and real-world studies—to summarize current knowledge on the efficacy and safety of venetoclax-based regimens in AML patients. Results: Recent studies have evaluated VEN-based regimens in newly diagnosed (ND) and refractory/relapsed (R/R) AML patients, showing the efficacy of these treatments. VEN with hypomethylating agents (HMAs) became the standard-of-care for elderly/unfit AML patients. Recent studies strongly support the effectiveness of VEN-based regimens in frontline treatment of adult AML patients eligible for intensive treatments. VEN-based therapies were also used in combination with targeted therapies, thus generating triplet therapeutic regimens that are under evaluation for the treatment of some AML subtypes. However, the response to VEN+HMAs is highly variable and in part depends on tumor genetics; some patients are resistant or relapse following VEN-based treatments and future studies will be required to develop therapeutic strategies able to circumvent resistance and to identify patients at high risk of relapse. Prospective randomized trials are required to establish the real efficacy of VEN in various clinical settings and to refine maintenance and discontinuation strategies, aiming to improve long-term outcomes and to make more safe treatments based on VEN. Full article

Background: The clinical efficacy of the BCL-2 inhibitor venetoclax in acute myeloid leukemia (AML) is significantly undermined by the frequent emergence of drug resistance, which precipitates disease progression and poor patient outcomes. However, the molecular landscape of this resistance remains insufficiently understood. Methods: To address this, we developed venetoclax-resistant AML cell models and utilized transcriptomic profiling integrated with comprehensive in vitro and in vivo functional assays. Results: Resistant cells demonstrated sustained proliferation even under the suppression of BCL-2, MCL-1, and key intrinsic apoptotic markers, including cleaved PARP and caspase-9, indicating a bypass mechanism independent of classical BCL-2 signaling. Compared to their sensitive counterparts, resistant Kasumi-1 (VENK) and MV4-11 (VENM) cells exhibit aggressive growth phenotypes in vitro and in vivo, characterized by larger, more numerous spheroids and colonies, alongside heightened tumorigenicity in murine models. Transcriptomic profiling and KEGG analysis identified the neuroactive ligand–receptor interaction (NLRI) pathway as a significant signaling node shared between these resistant lines. While multiple NLRI-associated genes were altered, CHRNB4 was consistently and significantly downregulated in both VENK and VENM cells and tumors. Re-expression of CHRNB4 in resistant cells, a primary gain-of-function approach, significantly impaired colony formation, and tumor growth in vivo. Clinically, CHRNB4 downregulation correlates with shortened overall survival and diminished response to venetoclax. Conclusions: Our findings implicate the NLRI pathway in venetoclax resistance and identify CHRNB4 as a robust prognostic indicator and a promising therapeutic target for developing next-generation AML strategies. Full article

Objectives: Allogeneic stem cell transplantation (HSCT) is curative in acute myeloid leukemia (AML) but is limited by relapse and non-relapse mortality (NRM). Metabolomic prognostic value is unclear. We assessed whether plasma metabolite profiles at diagnosis, pre-transplant, and post-transplant are associated with overall survival (OS) and cause-specific mortality. Methods: We retrospectively analyzed plasma metabolites from 63 AML patients undergoing HSCT (263 samples). Results: Higher levels of valine (hazard ratio [HR] 24.454), citrulline (HR 20.478), 5-oxoproline (HR 11.766), and glutamine (HR 8.701) associated with higher NRM, while inosine diphosphate (HR 0.091) and pyridoxamine-5′-phosphate (HR 0.313) associated with lower NRM. For relapse-related mortality (RRM), higher levels of phenylalanine (HR 26.585), leucine/isoleucine (HR 10.755), indolepyruvate (HR 7.676), and creatinine (HR 13.874) were associated with higher RRM, while trans-4-hydroxy-L-proline (HR 0.101) was associated with lower RRM. Higher post-transplant ornithine (HR 0.063), 3-sulfocatechol (HR 0.590), and indole-3-acetate (HR 0.359) were associated with improved OS. Mixed-effects modelling identified lower dehydroascorbate and citrate in relapsed patients, with dehydroascorbate remaining significant after false discovery rate adjustment. Conclusions: Metabolomic profiling nominated candidate metabolites for validation in larger prospective studies and elucidated mechanistic pathways, potentially informing novel interventions or risk-adapted monitoring strategies in HSCT. Full article

Background: A major technical challenge in single-cell transcriptomics is the absence of an integrative analytic pipeline that can simultaneously leverage gene regulatory network (GRN) architecture, AI-assisted gene panel discovery, and functional relevance analyses to generate coherent biological insights. Existing approaches often treat these components independently, focusing on clusters, marker genes, or predictive features without integrating them into a mechanistically grounded framework. Consequently, comprehensive screening that links regulatory association, gene signature screening, and functional interpretation within single-cell datasets remains limited, underscoring the need for an integrated strategy. Methods: We developed an integrative bioinformatics pipeline based on Gene regulatory network–AI–Functional Analysis (GAFA), combining latent-space integration, unsupervised clustering, diffusion pseudotime analysis, lineage-resolved generalized additive modeling, GRN inference, and machine learning-based gene panel discovery. This framework enables systematic mapping of cell-state structure, reconstruction of differentiation and effector trajectories, and identification of transcriptional and regulatory features strongly associated with clinical outcomes. As a case study, we applied the pipeline to NK cell transcriptomes from six CML patients (two early relapse, two late relapse, two durable treatment-free remission—TFR; 15 samples) collected at TKI discontinuation and 6–12 months after therapy cessation. Results: We reanalyzed publicly available scRNA-seq data from a previously published CML cohort to evaluate NK-cell transcriptional programs associated with treatment-free remission and relapse. We resolved six transcriptionally distinct NK cell states spanning CD56^bright-like cytokine-responsive, early activated, terminally mature, cytotoxic, lymphoid trafficking, and HLA-DR⁺ immunoregulatory populations, each exhibiting outcome-specific compositional differences. Pseudotime analysis revealed two major NK cell lineages—a maturation trajectory and a cytotoxic effector trajectory. TFR samples displayed balanced occupancy of both lineages, whereas early relapse samples showed marked depletion of the maturation branch and preferential accumulation in cytotoxic end states. AI-guided feature selection and random forest modeling identified an 18-gene panel that distinguished NK cells from TFR and relapse samples in an exploratory manner. Among them, CST7, FCER1G, GNLY, GZMA, and HLA-C were conventional NK-associated genes, whereas ACTB, CYBA, IFITM2, IFITM3, LYZ, MALAT1, MT2A, MYOM2, NFKBIA, PIM1, S100A8, S100B, and TSC22D3 were novel. The GRN inference further uncovered outcome-specific regulatory modules, with RUNX3, EOMES, ELK4, and REL regulons enriched in TFR, whereas FOSL2 and MAF regulons were enriched in relapse, and their downstream targets linked to IFN-γ signaling, metabolic reprogramming, and immunoregulatory feedback circuits. Conclusions: This AI-enabled single-cell analysis demonstrates how NK cell state composition, differentiation trajectories, and regulatory network rewiring collectively shape TFR versus relapse following TKI discontinuation in CML. The integrative pipeline provides a modular framework that could be extended to additional datasets for data-driven biomarker discovery and mechanistic stratification, and highlights candidate transcriptional regulators and NK cell programs that may be leveraged to improve remission durability, pending validation in larger patient cohorts. Full article