AML epigenetic therapies face network-driven resistance, review finds

Epigenetic therapies can correct disrupted gene regulation in acute myeloid leukemia, but most responses remain partial or temporary because leukemic cells reconfigure chromatin and transcriptional networks to survive treatment, according to a comprehensive review of established and emerging approaches. The analysis finds that durable control is unlikely to come from correcting a single molecular defect. Instead, it may require biomarker-selected combinations that target cooperating pathways, monitor epigenetic reprogramming at the leukemic stem-cell level and account for non-coding RNAs that may reflect or modulate specific regulatory states.

Acute myeloid leukemia is a heterogeneous blood cancer in which genetic alterations, including mutations in DNMT3A, TET2, ASXL1 and IDH1 or IDH2, interact with dynamic changes in DNA methylation, histone marks, higher-order chromatin organization and gene expression. This combination sustains the capacity of malignant cells to self-renew, resist differentiation and move between transcriptional states. Because many epigenetic changes are reversible, they offer attractive therapeutic targets. Yet the same plasticity can let residual cells adapt, allowing pre-leukemic clones to survive and disease to relapse. The review therefore treats AML not as a collection of isolated molecular defects, but as a system in which relatively stable mutations establish a permissive framework that is continually reshaped by epigenetic regulation.

Hypomethylating agents such as azacitidine and decitabine were the first epigenetic drugs incorporated into routine myeloid treatment. They trap the DNA methyltransferase DNMT1 and promote passive loss of methylation during cell division, partially restoring silenced genes associated with differentiation and tumor suppression. Their principal role is disease modification rather than rapid elimination of leukemic blasts. The addition of venetoclax substantially improved remission and survival in the phase III VIALE-A trial for newly diagnosed patients unsuitable for intensive chemotherapy, but median survival was still measured in months rather than years, and outcomes varied with mutations such as NPM1, IDH1/2 and TP53. Failure is often linked to incomplete demethylation, compensatory methylation regulators, persistent founder clones and resilient leukemic stem-cell compartments, even when bulk blast measurements suggest a response.

Mutant IDH1 and IDH2 offer a more genetically defined route into the same methylation machinery. These enzymes produce the oncometabolite R-2-hydroxyglutarate, which inhibits TET enzymes and JmjC-domain histone demethylases, producing widespread hypermethylation and impaired myeloid differentiation. Ivosidenib and enasidenib demonstrated that suppressing this metabolite can restore differentiation, although responses are often incomplete and relapse remains common. The phase III AGILE trial subsequently showed that ivosidenib plus azacitidine improved event-free and overall survival over azacitidine alone in newly diagnosed, intensive-therapy-ineligible IDH1-mutant AML. Resistance nevertheless emerges through inhibitor-binding mutations, switching to the other IDH isoform, activation of proliferative signaling, altered metabolism and expansion of pre-leukemic or IDH-wild-type clones. Correcting metabolism can therefore create a differentiation window without removing every survival route.

Menin inhibitors attack a different regulatory layer by disrupting the KMT2A-menin complex, which sustains HOXA9 and MEIS1 programs in KMT2A-rearranged and NPM1-mutated disease. Revumenib, an oral menin inhibitor, produced complete remission or complete remission with partial hematologic recovery in roughly one fifth of evaluable patients in the phase I/II AUGMENT-101 trials, with some responders proceeding to allogeneic transplantation. It also illustrates the gap between molecular correction and clinical control. MEN1 mutations can impair revumenib binding, while TP53 inactivation may blunt apoptosis even when HOX and MEIS targets are successfully downregulated. Other work indicates that loss of PRC1.1 components can reactivate noncanonical targets such as MYC. Combination regimens are consequently being evaluated with hypomethylating agents, venetoclax, intensive chemotherapy and other targeted drugs, but longer follow-up is needed to establish durable benefit.

Several other chromatin targets remain investigational rather than established treatments. The DOT1L inhibitor pinometostat strongly suppressed H3K79 methylation yet produced few objective responses, apparently because some resistant cells continued to rely on DOT1L as a physical scaffold within the KMT2A complex rather than on its catalytic activity. Broad histone deacetylase inhibition has similarly been limited by a narrow therapeutic window and toxicity, encouraging interest in selective agents such as HDAC6 inhibitors. LSD1 inhibitors can promote myeloid maturation in models, but early clinical data remain limited and adaptive chromatin co-dependencies may undermine monotherapy. BET inhibitors have produced occasional responses while causing myelosuppression and other toxicities; preclinical studies show that rapid redistribution of the P300 acetyltransferase can preserve oncogenic transcription after BET blockade. Newer targets include PRMT5 in spliceosome-mutant disease and KAT6A/KAT7 in NUP98-rearranged leukemia, for which early clinical or preclinical evidence supports further biomarker-guided combination studies.

Non-coding RNAs provide another, still exploratory layer within these networks. Most act after transcription by promoting messenger RNA degradation, repressing translation or acting as molecular sponges, but selected long non-coding RNAs have been proposed to interact directly with chromatin regulators. HOTTIP is elevated in KMT2A-rearranged and NPM1-mutated models and has been reported to interact with WDR5/KMT2A complexes, helping maintain active chromatin at the HOXA locus. Disrupting HOTTIP in preclinical models reduced HOXA transcription, impaired clonogenic growth and prolonged survival. HOXA-AS2 has separately been proposed to recruit EZH2 and PRC2 to targets including LATS2, reinforcing repressive H3K27me3 chromatin. These observations remain context-dependent, and the specificity of many lncRNA-chromatin interactions is disputed. The review therefore does not classify non-coding RNAs as a separate epigenetic axis comparable to DNA methylation or histone modification.

Clinical translation of non-coding RNA biology is constrained by variable expression, functional redundancy among RNA family members, inconsistent results across cohorts and differences in samples, platforms and normalization methods. Delivering oligonucleotides or small interfering RNAs directly into leukemia cells also presents obstacles involving delivery, toxicity and compensation by parallel regulatory circuits. Their nearer-term value may therefore lie in measurement rather than direct treatment. HOXA-associated RNAs such as HOTTIP and HOXBLINC could potentially indicate how completely menin inhibition has silenced HOX loci, while other non-coding RNAs may help classify residual disease or track differentiation. These remain hypotheses requiring prospective testing alongside genomic, methylation and chromatin readouts, not established clinical markers.

The review’s central conclusion is that resistance in AML extends beyond ordinary drug failure or secondary mutations. Leukemic cells can preserve core survival programs by changing cofactors, reactivating alternative transcriptional circuits and maintaining epigenetic states in stem or progenitor compartments. This explains why strong laboratory activity has rarely translated into durable benefit in unselected patients, and why indiscriminate combinations have sometimes added toxicity without improving survival. Near-term progress is more likely through trials selected by molecular dependencies, combinations that address cooperating pathways, schedules designed around predictable adaptation, and single-cell multi-omics that distinguish temporary blast reduction from genuine reprogramming. Epigenetic therapy remains clinically actionable in AML, but its future depends on treating the network and measuring the state of its cells, rather than assuming that correcting one regulator is enough.

Subject of Research: Network redundancy, resistance mechanisms and non-coding RNA modulation in epigenetic therapy for acute myeloid leukemia

Article Title: Epigenetic therapy in AML: network redundancy, resistance, and non-coding RNAs

Article References: Fernandes, M. D. C. R., Scandolara, T. B., Mota, J. A., Manso, M. P., Furtado, G. P., Maranhão, S. S., Pessoa, C. D. Ó., & Furtado, C. L. M. (2026). Epigenetic therapy in AML: network redundancy, resistance, and non-coding RNAs. Epigenetics Communications, 6(1), Article 14. https://doi.org/10.1186/s43682-026-00054-6

Image Credits: AI Generated

DOI: 10.1186/s43682-026-00054-6

Keywords: acute myeloid leukemia, epigenetic therapy, DNA methylation, chromatin regulation, menin inhibitors, IDH inhibitors, hypomethylating agents, non-coding RNAs, drug resistance, cancer genomics, Epigenetic, therapy