The KMT2A-r DNA methylation signature — focal hypermethylation, no global hypomethylation, and a hypomethylating agent that did not help
also called MLL-r methylation signature · promoter hypermethylation infant ALL · global hypomethylation KMT2A-r · epigenetic priming azacitidine
In infant leukemia with a KMT2A fusion, many gene switches get chemically locked shut by DNA methylation — and the more locked they are, the worse the outcome. But the genome as a whole is not under-methylated, and a drug that strips methylation off proved safe without improving survival.
The same thing, for a scientist
Three claims need separating. (1) Focal, promoter CpG-island hypermethylation is real and is the signature feature, but it is subtype-specific: t(4;11) and t(11;19) infant ALL are extensively hypermethylated while t(9;11) and KMT2A-germline infant ALL epigenetically resemble normal bone marrow. (2) A global hypomethylation phenotype does NOT exist — direct measurement of LINE-1, Alu and satellite-alpha plus LUMA in 45 KMT2A-r infant ALLs found no global hypomethylation and a moderate trend towards global hyper-methylation. (3) Superimposed on the hypermethylated background there is focal hypomethylation at specific loci, notably AP-1 (FOS/JUN) and RUNX factors, which is functionally exploitable. Clinically, azacitidine priming before chemotherapy in infants was safe and pharmacodynamically active but did not improve survival.
Focal promoter hypermethylation, and it splits by fusion partner2 deeper
The founding clinical observation is also the one most often flattened into 'KMT2A-r is hypermethylated'. What the data actually show is a split: KMT2A translocations t(4;11) and t(11;19) characterised extensively hypermethylated leukemias, whereas t(9;11)-positive infant ALL and KMT2A-germline infant ALL epigenetically resembled normal bone marrow. And within the hypermethylated group the degree of promoter hypermethylation influenced relapse-free survival: patients with accentuated methylation were at high relapse risk. Zebularine reversed the methylation and induced apoptosis 1.
So methylation is a partner-dependent, prognostically graded feature, not a uniform class marker. That is the same kind of intra-class heterogeneity as the HOXA-high/low split, and it should make anyone nervous about treating 'KMT2A-r' as one biological entity.
An independent cohort, using the HELP assay on 5 KMT2A-r infant leukemias versus 5 other childhood ALLs and 5 normals, found unsupervised clustering separated the groups and global hypermethylation in the KMT2A-r cohort, with ratios of significantly hypermethylated to hypomethylated CpGs of 1.7 (versus normals) and 2.9 (versus other ALLs); 5 of 7 assayed hypermethylated genes were relatively silenced, and decitabine was dose- and time-dependently cytotoxic in KMT2A-r but not KMT2A-wild-type lines with re-expression of 4 of 5 silenced genes 2. Direction consistent; n=5 per group, so power is weak.
The global hypomethylation claim is false, and this was tested directly
In most cancers, promoter hypermethylation travels with global loss of methylation in non-promoter regions, which drives genomic instability. The obvious inference — that KMT2A-r infant ALL is globally hypomethylated — was tested and refuted.
High-resolution bisulfite pyrosequencing of LINE-1, Alu and satellite-alpha, plus the LUminometric Methylation Assay, across 45 KMT2A-r infant ALL, 11 germline-KMT2A infant ALL, 11 non-infant BCP-ALL and 9 normal paediatric bone marrows found that KMT2A-rearranged infant ALL is not characterised by global hypomethylation, despite its characteristic promoter CpG hypermethylation. Instead there was a moderate trend towards global hyper-methylation, and the non-promoter methylated sequences also responded to demethylating agents 3.
This is an unusually clean negative result with a real cohort, and it should be recorded as a refuted claim in the graph. Repetitive-element surrogates plus LUMA are not WGBS, so the finding is a strong constraint rather than a final word — but the direction is opposite to the hypothesis, not merely non-significant.
Focal hypomethylation does exist, and it points at AP-1
Genome-wide methylome plus transcriptome from 69 uniformly treated (Interfant-99/06) infant B-ALL patients, KMT2A-r and non-KMT2A-r, showed that infant B-ALL methylome signatures map to chromatin states related to enhancer and transcriptional control in normal haematopoietic cells, segregate KMT2A-r from non-KMT2A-r, and identify coordinated enrichment of the AP-1 members FOS/JUN and RUNX factors, consistent with significant enrichment of hypomethylated CpGs in those genes. AP-1 inhibition or ablation impaired KMT2A-r growth in vitro and in PDX models 4.
The smaller, earlier version of the same idea: activation of particular proto-oncogenes in t(4;11) is mediated by promoter hypomethylation, and connectivity mapping on the most-hypomethylated genes nominated HDAC inhibitors, which killed primary t(4;11) infant ALL cells with downregulation of MYC, SET, RUNX1, RAN and the fusion product itself, and restored DNA methylation 5.
The synthesis: the methylome is hypermethylated on average, with a targeted set of oncogenic loci specifically stripped of methylation. The two observations are not in conflict; they describe different scales, and only the focal one is directly actionable. Note also the pleasing convergence with step 1: the CXXC domain's job is to protect specific CpGs from methylation, so focal hypomethylation at fusion-relevant loci is exactly what the fusion protein should produce.
The clinical test was run, and it was negative1 deeper
This is the most important paragraph in this node, because the biology above generated an obvious therapeutic idea and the idea was tested prospectively.
COG trial AALL15P1 gave 5 days of azacitidine as epigenetic priming before chemotherapy in four post-induction courses in infants with newly diagnosed KMT2A-rearranged ALL. It was safe (dose-limiting toxicity in only 2 of 31 evaluable patients, 6.5%) and pharmacodynamically active (whole-genome bisulfite sequencing of PBMCs showed decreased DNA methylation in 87% of samples tested). Event-free survival was similar to prior studies, and the authors state plainly that adding azacitidine did not impact survival 6.
Three readings, and the lab should hold all three:
- Deflationary. Promoter hypermethylation may be a passenger correlate of the cell of origin rather than a driver. Demethylating it changes the epigenome without changing the disease.
- Sequencing/scheduling. Priming before chemotherapy may be the wrong pairing; the mechanistically coherent combination is a demethylating agent with a menin inhibitor (which evicts the fusion complex), not with cytotoxics. That combination has not been reported.
- Wrong endpoint. Because methylation status is prognostic within t(4;11)/t(11;19) 1, an unstratified trial dilutes any effect. The trial did not enrich for the accentuated-methylation high-relapse-risk group in which the rationale is strongest.
A caution on framing pan-ALL methylation data: a 764-child, 435,941-CpG survey defined pan-ALL and subtype-specific signatures including 9,406 shared predominantly hypermethylated CpGs and 6,612 CpGs predominantly hypermethylated at relapse, with subtype-specific CpGs stratifying relapse-free survival 7 — but that paper reports no KMT2A-r-specific CpG counts, so its numbers must not be attributed to KMT2A-r.
Are de novo DNA methyltransferases required? Apparently not
If hypermethylation were a driver, the de novo methyltransferases should be required. They are not, in this fusion class. DNMT3A is dispensable for RUNX1-RUNX1T1- and MLL-AF9-driven self-renewal, whereas PML-RARA requires DNMT3A methyltransferase activity 8.
And the DNMT3B direction is counter-intuitive in a way that should give pause: inducible high Dnmt3b expression slowed leukemia development and impaired leukemia-stem-cell function; notably, MLL-AF9-induced leukemogenesis showed much less pronounced DNA hypermethylation on Dnmt3b expression than Myc-Bcl2 leukemias, yet was still delayed 9. More de novo methylation was anti-leukemic.
Taken together with the negative azacitidine trial, the honest position is: the KMT2A-r methylation signature is a robust diagnostic and prognostic biomarker with, so far, no demonstrated causal or therapeutic traction.
| chain | mechanism-kmt2a-fusion-causal-chain |
|---|---|
| signature | focal promoter CpG-island hypermethylation with globally normal-to-slightly-high methylation |
| global hypomethylation | false |
| global hypomethylation status | tested and refuted (n=45 KMT2A-r infant ALL; LINE-1/Alu/SAT-alpha pyrosequencing + LUMA) |
| hypermethylated tumour suppressor example | FHIT |
| prognostic | degree of hypermethylation within t(4;11)/t(11;19) stratifies relapse-free survival |
| dnmt requirement | DNMT3A dispensable for MLL-AF9 self-renewal; increased DNMT3B activity is anti-leukemic |
| causal status | robust biomarker; no demonstrated causal or therapeutic traction |
subtype split
[
{
"fusion": "fusion-kmt2a-aff1",
"methylation": "extensively hypermethylated"
},
{
"fusion": "fusion-kmt2a-mllt1",
"methylation": "extensively hypermethylated"
},
{
"fusion": "fusion-kmt2a-mllt3",
"methylation": "resembles normal bone marrow"
},
{
"fusion": "KMT2A-germline infant ALL",
"methylation": "resembles normal bone marrow"
}
]focal hypomethylation loci
[ "FOS", "JUN", "RUNX factors", "MYC", "SET", "RAN" ]
clinical trial
{
"nct": "NCT02828358",
"name": "COG AALL15P1",
"agent": "azacitidine priming",
"n_evaluable": 31,
"dlt_pct": 6.5,
"pd_active_pct": 87,
"survival_benefit": false
}- The one prospective clinical test of this biology (azacitidine priming, COG AALL15P1) was safe and on-target but did not improve survival, so the methylation signature may be a passenger of the cell of origin rather than a driver.
- Hypomethylating agents strip methylation genome-wide; since the CXXC domain's function is to protect specific CpGs from methylation, a demethylating agent could in principle broaden rather than narrow the fusion's accessible target set.
- Increased DNMT3B-mediated methylation is anti-leukemic in the mouse, so demethylating therapy may act against the grain of the biology.
- Methylation-based stratification risks mis-classifying t(9;11) and KMT2A-germline infant ALL, which look epigenetically normal despite aggressive disease.
- Would a hypomethylating agent combined with a menin inhibitor, rather than with chemotherapy, work? AALL15P1 tested the mechanistically weaker pairing. This is the most valuable untested combination arising from this node.
- Does baseline methylation status predict menin-inhibitor response? Methylation is prognostic within t(4;11)/t(11;19), and CXXC targeting depends on the methylome, so there is a specific mechanistic reason to expect it to.
- Has anyone profiled the KMT2A-r methylome by WGBS rather than repetitive-element surrogates? The global-hypomethylation refutation is directionally strong but methodologically indirect.
- Is the t(9;11) 'normal-like' methylome a partner effect or an age/cell-of-origin effect? t(9;11) infant ALL is rare and the confound with age is unresolved, exactly as for the HOXA split.
- Why is increased DNMT3B activity anti-leukemic in MLL-AF9? If de novo methylation opposes the fusion, the therapeutic vector may point towards methylating rather than demethylating agents — the opposite of current practice.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In infant ALL, KMT2A t(4;11) and t(11;19) characterise extensively promoter-hypermethylated leukemias whereas t(9;11)-positive and KMT2A-germline infant ALL epigenetically resemble normal bone marrow; among t(4;11)/t(11;19) patients the degree of promoter hypermethylation influenced relapse-free survival, with accentuated methylation conferring high relapse risk, and zebularine reversed methylation and induced apoptosis.
Specific promoter methylation identifies different subgroups of MLL-rearranged infant acute lymphoblastic leukemia, influences clinical outcome, and provides therapeutic options · Stumpel DJPM, Schneider P, van Roon EHJ, et al. (Stam RW, senior author) · Blood 114(27):5490-5498 2009What this evidence cannot show
Differential methylation hybridisation, pre-450k-array technology; the abstract does not report CpG counts or per-subgroup cohort sizes, and the survival analysis is a retrospective subgroup finding in a rare disease. t(9;11) infant ALL is uncommon, so the 'resembles normal marrow' group is small.
Evidence that argues against this (1)
- Promoter hypermethylation is not specific to KMT2A-rearranged disease. Nordlund et al., surveying 435,941 CpGs in 764 children with ALL, found 9,406 predominantly hypermethylated CpGs shared across ALL regardless of cytogenetic subtype. A signature that is present in the comparison group cannot by itself mark the KMT2A-r class, and this entry's E7 carries the same finding as support - the two readings need to be held together. Nordlund J et al. 2013
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
HELP-assay profiling of 5 KMT2A-rearranged infant leukemias versus 5 other childhood ALLs and 5 normals showed global hypermethylation in the KMT2A-r cohort with hypermethylated:hypomethylated CpG ratios of 1.7 and 2.9 respectively (p<0.001); 5 of 7 assayed hypermethylated genes were relatively silenced, and decitabine was cytotoxic in KMT2A-r but not KMT2A-wild-type lines with re-expression of 4 of 5 silenced genes.
Promoter hypermethylation in MLL-r infant acute lymphoblastic leukemia: biology and therapeutic targeting · Schafer E, Irizarry R, Negi S, et al. (Brown P, senior author) · Blood 115(23):4798-4809 2010What this evidence cannot show
n=5 per group — the direction is consistent with larger cohorts but the study is badly underpowered on its own. Note also an attribution correction: Figueroa ME is a co-author here, not a first author; any citation of 'Figueroa et al.' for infant-ALL methylation is a misattribution and should be corrected to this paper.
Evidence that argues against this (1)
- Bulk methylation profiles of leukaemic marrow are mixtures. Houseman et al. showed that DNA methylation signatures are themselves surrogate measures of leukocyte composition and that differences between case and control groups can be produced by a shift in cell mixture alone. A five-versus-five HELP comparison of KMT2A-r infant marrow against other childhood ALL and normal marrow differs in blast percentage and in lineage composition as well as in genotype, and no deconvolution was applied, so an unknown fraction of the reported global hypermethylation is composition rather than disease biology. Houseman EA et al. 2012
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Direct measurement of global methylation by bisulfite pyrosequencing of LINE-1, Alu and satellite-alpha plus the LUminometric Methylation Assay in 45 KMT2A-rearranged infant ALL, 11 germline-KMT2A infant ALL, 11 non-infant BCP-ALL and 9 normal paediatric bone marrows showed that KMT2A-rearranged infant ALL is NOT characterised by global hypomethylation despite its promoter CpG hypermethylation; instead there was a moderate trend towards global hypermethylation, and the non-promoter methylated sequences also responded to demethylating agents.
Absence of global hypomethylation in promoter hypermethylated Mixed Lineage Leukaemia-rearranged infant acute lymphoblastic leukaemia · Stumpel DJPM, Schneider P, van Roon EHJ, Pieters R, Stam RW · European Journal of Cancer 49(1):175-84 2013What this evidence cannot show
Repetitive-element surrogates (LINE-1/Alu/SAT-alpha) and LUMA rather than whole-genome bisulfite sequencing, so focal or compartment-specific hypomethylation would be missed. The result is nonetheless a directional refutation, not merely a null.
Evidence that argues against this (1)
- Repeat-element and global assays measure bulk methylation in bulk cells, so they are sensitive to blast percentage and to the composition of the normal marrow comparators. Nine normal paediatric marrows against 45 rearranged cases is a comparison between a homogeneous leukaemic population and a mixed normal one, which shifts global methylation independently of disease. Oncofusion knowledge graph (internal analysis) 2026
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Genome-wide methylome and transcriptome analysis of 69 uniformly treated (Interfant-99/06) infant B-ALL patients showed methylome signatures mapping to enhancer and transcriptional-control chromatin states in normal haematopoietic cells, segregating KMT2A-r from non-KMT2A-r cases, with coordinated enrichment of AP-1 members FOS/JUN and RUNX factors consistent with significant enrichment of hypomethylated CpGs in those genes; AP-1 inhibition or ablation impaired KMT2A-r growth in vitro and in PDX models.
Integrative methylome-transcriptome analysis unravels cancer cell vulnerabilities in infant MLL-rearranged B cell acute lymphoblastic leukemia · Tejedor JR, Bueno C, Vinyoles M, et al. (Menendez P, senior author) · Journal of Clinical Investigation 131(13):e138833 2021What this evidence cannot show
Array-based methylome; the AP-1 dependency is established pharmacologically and by ablation but the causal link from hypomethylation to AP-1 activity is correlative. AP-1 is broadly active in normal cells, so the therapeutic index of AP-1 inhibition is unaddressed.
Evidence that argues against this (2)
- The normal reference is the weak point of a cell-of-origin methylome argument. Mapping infant KMT2A-r methylomes onto chromatin states of *normal haematopoietic* cells assumes the right normal comparator is postnatal haematopoiesis; infant KMT2A-r ALL initiates in utero. Combined with the cell-mixture problem - bulk methylation differences track leukocyte composition (Houseman 2012) - the direction of causation between methylome state and fusion activity is not resolved by this design. Houseman EA et al. 2012
- Cross-cohort methylome comparisons are exposed to batch structure. Leek et al. document that technical batch effects in high-throughput data are widespread, frequently correlate with the outcome of interest, and have produced incorrect conclusions; an Interfant-99/06 case series profiled separately from its normal references carries exactly that risk. Leek JT et al. 2010
- E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Activation of particular proto-oncogenes in t(4;11) infant ALL is mediated by promoter hypomethylation; connectivity mapping on the genes most significantly hypomethylated versus healthy bone marrow nominated HDAC inhibitors, which killed primary t(4;11) infant ALL cells with downregulation of MYC, SET, RUNX1, RAN and the KMT2A::AFF1 fusion product, and restored DNA methylation.
Connectivity mapping identifies HDAC inhibitors for the treatment of t(4;11)-positive infant acute lymphoblastic leukemia · Stumpel DJPM, Schneider P, Seslija L, et al. (Stam RW, senior author) · Leukemia 26(4):682-692 2012What this evidence cannot show
Connectivity mapping is a hypothesis-generating signature-matching method; HDAC inhibitors have pleiotropic effects and 'restored DNA methylation' is a correlated observation, not a demonstrated mechanism of their cytotoxicity. Ex-vivo primary cells only.
Evidence that argues against this (1)
- Connectivity mapping nominates compounds whose expression signatures oppose an input signature; it does not establish that the drug acts on the mechanism that generated it. HDAC inhibitors kill primary leukaemia cells broadly, and MYC downregulation is among the most common consequences of HDAC inhibition in any cell — so the kill is weak confirmation of a hypomethylation-driven mechanism. Oncofusion knowledge graph (internal analysis) 2026
- E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In COG trial AALL15P1, 5 days of azacitidine as epigenetic priming before chemotherapy in four post-induction courses in infants with newly diagnosed KMT2A-rearranged ALL was safe (dose-limiting toxicity in 2 of 31 evaluable patients, 6.5%) and pharmacodynamically active (decreased DNA methylation by WGBS in 87% of samples tested), but event-free survival was similar to prior studies and azacitidine did not impact survival.
Azacitidine as epigenetic priming for chemotherapy is safe and well-tolerated in infants with newly diagnosed KMT2A-rearranged acute lymphoblastic leukemia: Children's Oncology Group trial AALL15P1 · Guest EM, Kairalla JA, Devidas M, et al. (Brown PA, senior author) · Haematologica 109(12):3918-3927 2024What this evidence cannot show
A single-arm phase 2 safety/feasibility study with historical comparison, not a randomised efficacy trial, so 'did not impact survival' is a comparison to prior studies rather than a controlled result. Patients were not stratified by baseline methylation status, which is the group in which the rationale is strongest. Azacitidine was paired with chemotherapy, not with a menin inhibitor.
Evidence that argues against this (1)
- The therapeutic corollary of the signature has now been tested and did not hold. AALL15P1 showed azacitidine priming was pharmacodynamically on target - decreased methylation by WGBS in 87% of samples - yet event-free survival was similar to prior studies. Demethylation was achieved and survival did not move, which argues that the hypermethylated state is not by itself the load-bearing dependency this mechanism entry implies. Guest EM et al. 2024
- E7established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
A survey of 435,941 CpG sites in 764 children with ALL at diagnosis and 27 at relapse defined four signatures: 9,406 shared predominantly hypermethylated CpGs across ALL regardless of cytogenetics; subtype-unique hyper- and hypomethylated sets differing in enrichment at active versus repressed chromatin; subtype-specific promoter and enhancer differential methylation correlated with expression; and 6,612 CpGs predominantly hypermethylated at relapse, with subtype-specific CpGs stratifying relapse-free survival.
Genome-wide signatures of differential DNA methylation in pediatric acute lymphoblastic leukemia · Nordlund J, Backlin CL, Wahlberg P, et al. (Syvanen AC, senior author) · Genome Biology 14(9):r105 2013What this evidence cannot show
These are pan-ALL and subtype-level numbers; the abstract reports NO KMT2A-r-specific CpG counts, so none of these figures may be attributed to KMT2A-r. KMT2A-r cases are a small minority of a predominantly non-infant pediatric ALL cohort.
Evidence that argues against this (1)
- This is a pan-ALL survey and the KMT2A contribution is one subtype among many. Four signatures derived across 435,941 CpGs in 764 children carry a large multiple-testing burden, and the 27-sample relapse arm differs from diagnosis in treatment exposure and blast purity as well as in methylation. Leukemia 2021
- E8strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
DNMT3A is dispensable for RUNX1-RUNX1T1- and MLL-AF9-driven self-renewal, whereas PML-RARA requires DNMT3A methyltransferase activity to initiate acute promyelocytic leukemia — arguing that de novo DNA methylation is not a required driver mechanism in KMT2A-fusion leukemia.
PML-RARA requires DNA methyltransferase 3A to initiate acute promyelocytic leukemia · Cole CB, Verdoni AM, Ketkar S, et al. (Ley TJ, senior author) · Journal of Clinical Investigation 126(1):85-98 2016What this evidence cannot show
Mouse retroviral/ex-vivo self-renewal plus transplantation; MLL-AF9 in a myeloid context, not KMT2A::AFF1 in a lymphoid one, and DNMT3B was not deleted in the same experiment so paralogue compensation is not excluded.
Evidence that argues against this (1)
- DNMT3A dispensability is the entry's own strongest argument against a methylation-driven mechanism, and it is a mouse result in MLL-AF9 myeloid leukaemia. It bounds de novo methylation as a REQUIREMENT for initiation and leaves open both maintenance methylation through DNMT1 and a role in human B-lymphoid disease. Oncofusion knowledge graph (internal analysis) 2026
- E9moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Tetracycline-inducible high Dnmt3b expression slowed leukemia development in serial transplantation and impaired leukemia-stem-cell function; MLL-AF9-induced leukemogenesis showed much less pronounced DNA hypermethylation upon Dnmt3b expression than Myc-Bcl2 leukemias yet was still delayed, and AML patients with high DNMT3B-target expression had inferior survival.
Increased DNA methylation of Dnmt3b targets impairs leukemogenesis · Schulze I, Rohde C, Scheller-Wendorff M, et al. (Muller-Tidow C, senior author) · Blood 127(12):1575-1586 2016What this evidence cannot show
Murine, with supraphysiological induced Dnmt3b expression; the direction (more methylation is anti-leukemic) is the opposite of the therapeutic hypothesis and complicates any simple hypermethylation-as-driver model, but overexpression systems can produce non-physiological methylation distributions.
Evidence that argues against this (1)
- Forced Dnmt3b overexpression from a tetracycline promoter is a gain-of-function far above physiological, and the informative comparison — much less hypermethylation in MLL-AF9 than in Myc-Bcl2 leukaemias yet still delayed — points to a methylation-independent effect of the transgene rather than to methylation as the mechanism. Oncofusion knowledge graph (internal analysis) 2026
- ← analogous toPentimento and restoration — the fusion overpaints the cell, and cleaning reveals whatever was underneath
the methylome as the ground beneath the fusion overpaint
competing primary epigenetic explanation raised by the counter-evidence pass: protection of specific CpGs from DNA methylation at MLL-fusion target loci is causally required for leukaemogenesis, while locus targeting itself is preserved even by DNA-binding-dead domain swaps
t(4;11) and t(11;19) are extensively promoter-hypermethylated; t(9;11) is not
EM-seq arm tests methylation as the candidate memory carrier
Counter-evidence (Guest 2024, COG AALL15P1, PMID 38867582): azacitidine priming before chemotherapy in infants with newly diagnosed KMT2A-r ALL was well tolerated and demethylated 87% of tested samples by WGBS, but event-free survival matched prior studies and the authors state azacitidine did not impact survival - the chemosensitisation rationale is refuted in this exact population
Counter-evidence (Guest 2024, COG AALL15P1, PMID 38867582): azacitidine priming in infants with newly diagnosed KMT2A-r ALL achieved demethylation (WGBS, 87% of tested samples) with 2/31 dose-limiting toxicities but produced event-free survival similar to prior studies and no survival impact - the demethylation therapy this methylation literature motivates has been tested unselected and failed
- confers resistance to →Revumenib (SNDX-5613, Revuforj) — first approved menin–KMT2A interface inhibitor
a second, menin-independent silencing layer
Counter-evidence: COG AALL15P1 (PMID 38867582) achieved on-target demethylation in 87% of infants with KMT2A-r ALL and produced no survival benefit, and bulk methylome differences are confounded by blast/leukocyte composition (Houseman 2012, PMID 22568884).
EVI1-positive MLL-rearranged leukaemias carry globally elevated 5mC, mirroring HSC-derived versus GMP-derived murine leukaemias
The KMT2A-rearranged methylation signature is not explained by TET2 co-mutation, which is depleted in this disease.
A candidate genetic cause of hypermethylation in KMT2A-rearranged disease that the existing signature entry does not consider.
The therapeutic arm of the promoter-hypermethylation finding: decitabine re-expressed 4 of 5 silenced genes with MSP-confirmed methylation loss.
The clinical test of the hypermethylation signature as a therapeutic target: the methylome moved and the survival curve did not.
A genetic route to the hypermethylation the signature entry describes, which that entry does not currently consider.
modifier, not a step - the methylome the CpG address reads
wild-type allele silencing is reversed by DNMT and HDAC inhibitors
azacitidine priming; on-target demethylation in 87% of samples, no survival effect