Hypothesis: t(4;11) infant ALL is a hypermethylated genome whose transposable elements can be reawakened, and that reawakening is an untested immunotherapeutic lever
also called TE reactivation KMT2A-r · viral mimicry infant ALL · ERV derepression t(4;11)
Our DNA is full of ancient virus-like sequences that are normally kept switched off by chemical tags. t(4;11) infant leukemia is unusually heavily tagged, and drugs that strip the tags wake these sequences up in other cancers and make the cell look infected to the immune system. A trial has now shown those tags can be stripped safely in these infants - but doing it alongside chemotherapy did not help them live longer, and nobody has yet checked whether the immune alarm even goes off.
The same thing, for a scientist
Two facts sit next to each other and have never been joined. First, t(4;11) and t(11;19) infant ALL are **extensively promoter-CpG-hypermethylated** relative to t(9;11) and KMT2A-wild-type infant ALL, the degree of hypermethylation tracks relapse-free survival, and the demethylating agent zebularine reverses methylation and induces apoptosis in these cells [E1]; the same group later showed by LINE-1/Alu/satellite-alpha pyrosequencing that these leukaemias also lack the global repeat hypomethylation typical of cancer and trend toward global hypermethylation, with repeats responsive to demethylating agents [E1 counter, PMID 22921182]. Second, in AML, decitabine globally reactivates multiple transposable elements, significantly derepresses ERV3-1 with promoter demethylation and increased chromatin accessibility and H3K4me3, and activates the double-stranded-RNA sensor RIG-I and IRF7 - the canonical viral-mimicry response - with superior in-vivo activity over cytarabine in monosomal-karyotype xenografts [E2]. Nobody has measured TE or ERV expression, dsRNA sensing or interferon output in a KMT2A-rearranged leukemia, before or after hypomethylating-agent exposure. The separate question of whether TE insertion polymorphisms contribute to leukemia risk has only weak, exploratory support [E3]. **Two corrections, cycle 35, that change what this hypothesis is for.** (i) The cytotoxicity-enhancement rationale is dead in this disease: COG AALL15P1 gave azacitidine priming before chemotherapy to infants with newly diagnosed KMT2A-r ALL, confirmed demethylation by WGBS in 87% of tested samples, and found event-free survival similar to prior studies, with the authors stating that azacitidine did not impact survival [E1 counter, PMID 38867582]. (ii) Viral mimicry is a class effect of hypomethylating agents demonstrated first in colorectal cancer [E2 counter, PMID 26317465], so its occurrence in a KMT2A-r blast would say nothing about the fusion. What survives is narrow and still worth testing: that the *magnitude* of derepression and interferon output is larger in the hypermethylated t(4;11)/t(11;19) subgroup than in methylation-normal infant ALL, and that the useful combination partner is an immunotherapy rather than more chemotherapy.
The specific claim, and why it is a therapeutic question not a curiosity1 deeper
The claim is narrow, and cycle 35 made it narrower: a hypomethylating agent in t(4;11) infant ALL should produce viral mimicry, more strongly than in methylation-normal infant ALL subtypes, and that immune consequence - not tumour-suppressor reactivation and not enhanced chemosensitivity - is the only remaining route by which an HMA could help in this disease.
The chemosensitisation rationale has been tested and failed. Children's Oncology Group AALL15P1 took exactly the gene-reactivation logic of E1 into the clinic: 5 days of azacitidine immediately before chemotherapy, in four post-induction courses, in infants with newly diagnosed KMT2A-r ALL. Safety was fine (2 of 31 evaluable patients with dose-limiting toxicity) and the pharmacodynamics worked - WGBS showed decreased methylation in 87% of samples tested. Event-free survival was similar to prior studies and the authors report plainly that azacitidine did not impact survival [E1 counter-evidence]. This is not an ambiguous negative: the drug reached its target, and nothing followed. Any future HMA proposal in this disease must explain what it is doing differently.
And the mimicry effect, if found, will not be fusion-specific. Viral mimicry was defined in colorectal cancer and the MDA5/MAVS/IRF7 dependence was established there [E2 counter-evidence]. Finding ERV derepression and an interferon signature in an azacitidine-treated KMT2A-r blast would demonstrate a property of the drug. Only the comparative prediction - larger effect in the hypermethylated subgroup - has fusion-relevant content, and it is cheap: existing infant ALL transcriptomes re-aligned with a TE-aware pipeline, stratified by methylation subgroup.
If the dominant pharmacology is interferon induction and antigen visibility, the right combination partner is an immunotherapy (blinatumomab, CAR-T, checkpoint blockade) rather than more chemotherapy, and the right biomarker is an interferon signature rather than a methylation score. AALL15P1 tested the chemotherapy pairing. Nobody has tested the immunotherapy pairing.
It also connects to a real weakness of the menin-interface strategy: if the fusion program is switched off by revumenib, blasts survive in a differentiating or quiescent state. Making those residual cells immunologically visible is a mechanistically independent second hit that resistance mutations in MEN1 cannot defeat.
Why an infant is the hardest place to try this
Genome-wide demethylation in a child under one year old acts during a period of active haematopoietic, neural and somatic development. TE derepression is exactly the class of effect that could have consequences invisible in a 6-month trial: new insertions, imprinting disturbance (see hyp-imprinted-loci-kmt2a-r), and chronic interferon signalling in a developing immune system. AALL15P1 partially answers the short-term safety question - 5-day azacitidine priming was well tolerated with 2/31 dose-limiting toxicities and confirmed demethylation - but it does not answer this one, because nobody measured TE output, interferon signature, insertion burden or imprinting in those patients, and the follow-up is far too short for a late developmental effect. The trial makes the safety case for a mechanistic correlative study; it does not make the case that there is nothing to look for.
Honest grading
- Direct KMT2A-r evidence for TE reactivation: none.
- Direct KMT2A-r evidence for the enabling condition (hypermethylation, HMA sensitivity in vitro): strong 1.
- Indirect leukemia evidence for the mechanism (HMA to TE reactivation to RIG-I/IRF7): good, in AML 2.
- TE insertion polymorphism as a risk factor: weak and exploratory — a bioinformatic association of 101 HERV-K (HML-2) insertion polymorphisms with AML at p < 0.05 with no replication cohort and no functional follow-up 3. We include it explicitly graded weak so that nobody later cites it as established.
| statement | The hypermethylated genome of t(4;11) infant ALL silences transposable elements as well as genes (supported: repeats are hypermethylated, not hypomethylated, in MLL-r infant ALL, PMID 22921182). Hypomethylating-agent exposure will reactivate TEs and endogenous retroviruses and trigger dsRNA sensing and an interferon response - which is a class effect of the drug, not a fusion property - and the fusion-relevant, still untested claim is that the magnitude of this response is greater in the hypermethylated t(4;11)/t(11;19) subgroup than in methylation-normal infant ALL, making an HMA plus immunotherapy the rational pairing. The HMA plus chemotherapy pairing has been tested in this exact population and did not improve survival (PMID 38867582). |
|---|---|
| origin | literature-gap |
| prior probability | 0.4 |
| planned experiments note | Moved out of a reference field on 2026-09-19. It named a node that does not exist, which made the pointer read as a real cross-reference and, for hypotheses, silently satisfied the 'needs at least one experiment or analysis' rule. The intent is preserved here; the node still needs writing. |
predictions
[ "TE/ERV transcript levels are lower in t(4;11) infant ALL blasts than in t(9;11) or KMT2A-wild-type infant ALL, matching the methylation gradient", "Azacitidine or decitabine in KMT2A::AFF1 lines induces ERV transcripts, cytoplasmic dsRNA, RIG-I/MDA5 activation and an IRF7-dependent interferon signature", "HMA pre-treatment increases MHC class I and antigen-presentation gene expression and improves killing by blinatumomab or CD19 CAR-T in vitro", "The interferon signature, not the global methylation change, predicts HMA response in existing infant ALL trial samples" ]
analyses
[ "ana-noncoding-gap-ledger-kmt2a-r" ]
falsifiers
[ "RNA-seq of HMA-treated KMT2A::AFF1 cells shows no significant TE or ERV derepression and no interferon-stimulated gene induction", "TE/ERV derepression and interferon output on HMA are no GREATER in t(4;11)/t(11;19) blasts than in methylation-normal infant ALL subtypes - which would remove the only fusion-relevant content, since derepression itself is a class effect of the drug", "TE expression in t(4;11) blasts is no lower than in methylation-normal infant ALL subtypes, removing the premise that they are silenced", "HMA pre-treatment does not improve immune-mediated killing in vitro, which would remove the therapeutic rationale even if derepression occurs", "Reanalysis of existing infant ALL trial transcriptomes shows no relationship between interferon signature and HMA benefit" ]
evidence grade by class
{
"direct_kmt2a_r_te_reactivation": "none",
"direct_kmt2a_r_promoter_hypermethylation": "strong",
"direct_kmt2a_r_repeat_element_hypermethylation": "moderate (LINE-1/Alu/SAT-alpha pyrosequencing, n=45, same group, PMID 22921182)",
"indirect_leukemia_mechanism": "strong (adult AML)",
"viral_mimicry_is_fusion_specific": "no - class effect of hypomethylating agents, defined in colorectal cancer (PMID 26317465)",
"clinical_test_of_hma_plus_chemotherapy_in_this_disease": "negative (PMID 38867582)",
"te_insertion_risk": "weak"
}prior probability history
[
{
"cycle": 1,
"value": 0.5,
"basis": "strong hypermethylation data plus strong AML viral-mimicry mechanism"
},
{
"cycle": 35,
"value": 0.4,
"basis": "the derepression prediction is now near-certain because it is a class effect (PMID 26317465), but the therapeutic content of the hypothesis was halved: the chemosensitisation arm is refuted in this exact population (COG AALL15P1, PMID 38867582) and only the immunotherapy-combination arm survives."
}
]falsifiers already met
[
{
"falsifier": "adding an HMA to chemotherapy improves outcome in infant KMT2A-r ALL",
"outcome": "met - refuted",
"source": "COG AALL15P1, PMID 38867582: demethylation confirmed by WGBS in 87% of tested samples, EFS similar to prior studies, azacitidine did not impact survival"
}
]corrections
[
{
"cycle": 35,
"date": "2026-09-19",
"withdrawn": "the framing that hypomethylating agents are an untested lever in this disease, and the implicit chemosensitisation rationale inherited from E1's title",
"replaced_with": "azacitidine priming has been tested in infant KMT2A-r ALL (COG AALL15P1), achieved demethylation, and did not improve survival; only the HMA-plus-immunotherapy arm remains open. Repeat-element methylation in MLL-r infant ALL has also been measured and is elevated, not reduced.",
"source": "PMID 38867582; PMID 22921182"
}
]existing graph nodes to read first
[ "trial-cog-aall15p1", "mechanism-kmt2a-r-dna-methylation-signature" ]
planned experiments
[ "exp-noncoding-dependency-crispr-screen-kmt2a-aff1" ]
- Genome-wide demethylation in an infant risks imprinting disturbance and developmental effects that no short trial can detect, and transposon mobilisation is the mechanistically worst case.
- Chronic interferon signalling suppresses haematopoietic stem cell function, so viral mimicry could impair marrow recovery in exactly the patients least able to tolerate it.
- If the effect is immune-mediated, it will be invisible in standard cell-line cytotoxicity assays and the strategy could be discarded for the wrong reason.
- Interferon induction can also upregulate inhibitory checkpoints, so a mimicry strategy could paradoxically blunt an adjacent immunotherapy unless checkpoint blockade is included.
- The obvious version of this strategy has already been run and was negative on survival, so a new HMA proposal in infant KMT2A-r ALL starts from a failed precedent and needs a mechanistic read-out built in from day one, not a second efficacy bet.
- Has anyone quantified TE or ERV expression in any KMT2A-rearranged leukemia dataset? Existing infant ALL RNA-seq could be re-aligned with a TE-aware pipeline at zero experimental cost.
- COG AALL15P1 (NCT02828358) banked serial samples from azacitidine-primed infants with KMT2A-r ALL and already has WGBS on them. Can the same material be re-sequenced with a TE-aware pipeline to ask whether ERVs were derepressed and whether an interferon signature appeared? This is the single highest-value experiment in this node and requires no new patients.
- Does menin inhibition change DNA methylation or TE expression at all? If revumenib leaves the hypermethylated state intact, HMA combination is mechanistically non-overlapping.
- Could TE reactivation generate new insertions in a fetal-origin leukemia, and is that risk measurable in treated patients rather than assumed absent?
- Since demethylation was achieved in AALL15P1 and survival was unchanged, was the interferon arm simply not engaged, or was it engaged and irrelevant without an immune effector? The two answers point to opposite next trials.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
t(4;11) and t(11;19) infant ALL are extensively promoter-CpG-hypermethylated whereas t(9;11) and KMT2A-wild-type infant ALL resemble normal bone marrow; degree of hypermethylation influences relapse-free survival, and zebularine reverses aberrant methylation and induces apoptosis in MLL-rearranged ALL cells.
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, Boer JM, de Lorenzo P, Valsecchi MG, de Menezes RX, Pieters R, Stam RW · Blood 2009What this evidence cannot show
Array-based differential methylation hybridisation with limited CpG coverage and small subtype groups; the paper says nothing about transposable elements, which are outside its probe design.
Evidence that argues against this (2)
- The therapeutic half of E1's own title - 'provides therapeutic options' - has now been tested in exactly this population and did not pay off. Children's Oncology Group AALL15P1 gave 5 days of azacitidine priming immediately before chemotherapy in four post-induction courses to infants with newly diagnosed KMT2A-r ALL. It was well tolerated (2/31 evaluable patients with dose-limiting toxicity), and whole-genome bisulfite sequencing confirmed decreased DNA methylation in 87% of samples tested - so the pharmacodynamic premise held. But event-free survival was similar to prior studies of newly diagnosed infant ALL and the authors state that incorporating azacitidine to enhance cytotoxicity did not impact survival. Demethylation in this disease is achievable and, added to chemotherapy, does not help. That refutes the cytotoxicity-enhancement rationale outright and leaves only the immune-combination rationale standing. Guest EM et al. 2024
- This corrects a claim of absence rather than the biology, and it cuts against a counter-argument this node might otherwise have been given. Promoter CpG island hypermethylation in cancer is usually accompanied by global hypomethylation of repeats - but the same group measured LINE-1, Alu and satellite-alpha methylation by bisulfite pyrosequencing plus LUMA in 45 MLL-rearranged infant ALL samples against germline-MLL infant ALL, non-infant BCP-ALL and normal paediatric marrow, and found *no* global hypomethylation. Instead they report a moderate trend toward global HYPERmethylation, and show those non-promoter sequences are themselves responsive to demethylating agents. So the node's premise that repeat elements are methylated in t(4;11) is directly supported, not merely inferred - but the node's framing that repeat-element methylation in this disease has never been measured is wrong and has been withdrawn. Same laboratory as E1, so not independent, and methylation is not expression. Stumpel DJPM et al. 2013
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In AML, decitabine globally reactivates multiple transposable elements and significantly derepresses ERV3-1 with promoter demethylation, increased chromatin accessibility and H3K4me3, activates the dsRNA sensor RIG-I and IRF7, with ERV3-1 and RIG-I induction confirmed in serially sorted patient blasts and superior survival versus cytarabine in patient-derived monosomal-karyotype xenografts.
Decitabine Induces Gene Derepression on Monosomic Chromosomes: In Vitro and In Vivo Effects in Adverse-Risk Cytogenetics AML · Greve G, Schüler J, Grüning BA, Berberich B, Stomper J, Zimmer D, Gutenkunst L, Bönisch U, Meier R, Blagitko-Dorfs N, Grishina O, Pfeifer D, Weichenhan D, Plass C, Lübbert M · Cancer Research 2021What this evidence cannot show
Adult AML with monosomal karyotype, a genotype at the opposite end of the mutation-burden spectrum from infant KMT2A-r ALL; the framing is hemizygous tumour-suppressor derepression, with TE/ERV reactivation as an accompanying observation. Myeloid, not B-lineage.
Evidence that argues against this (1)
- Viral mimicry is a class effect of DNA-demethylating agents, not a property of any particular leukaemia genotype. It was defined in colorectal cancer: low-dose 5-AZA-CdR induces dsRNAs derived at least in part from endogenous retroviral elements, activates the MDA5/MAVS/IRF7 pathway, and knocking down MDA5, MAVS or IRF7 individually abolishes the drug's effect on cancer-initiating cells. Observing ERV derepression and an interferon signature in a KMT2A-rearranged blast after azacitidine would therefore be a demonstration of what the drug does everywhere, not of anything the fusion does. The fusion-specific content of this hypothesis is limited to the claim that the *magnitude* of the effect is larger in the hypermethylated t(4;11) subgroup than in methylation-normal infant ALL subtypes - which is prediction 1, and is the only part worth an experiment. Roulois D et al. 2015
- E3weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper
A whole-genome-sequencing analysis reported 101 HERV-K (HML-2) transposable-element insertion polymorphisms statistically correlated with AML (chi-square p < 0.05), most of them in the 8q24.13-8q24.21 region near MYC, PVT1, CCDC26 and TRIB1.
HERV-K (HML-2) insertion polymorphisms in the 8q24.13 region and their potential etiological associations with acute myeloid leukemia · Camargo-Forero N, Orozco-Arias S, Perez Agudelo JM, Guyot R · Archives of Virology 2023What this evidence cannot show
Exploratory bioinformatics with no replication cohort, no correction described for the multiplicity implied by 101 reported associations, no functional follow-up, and the authors themselves state the relationship remains unclear. Included so the record shows it was considered and graded weak; it should not be cited as evidence of causation.
Evidence that argues against this (1)
- Nothing has replicated the HERV-K (HML-2) insertion-polymorphism association, and the one systematic clinical intervention on the methylation axis in this disease since - COG AALL15P1, which demethylated infants with KMT2A-r ALL and banked serial samples - did not examine TE insertions at all. Cited here only to record that the obvious opportunity to test whether HMA exposure mobilises transposable elements in these children existed and was not taken; it is not evidence for or against the association, which remains graded weak. Guest EM et al. 2024
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
- related to →Alu repeats and nuclear-matrix anchors in the KMT2A breakpoint region — the genome-feature side of the break
Correction (Stumpel 2013, PMID 22921182): LINE-1, Alu and satellite-alpha methylation was measured in 45 MLL-rearranged infant ALL samples and shows NO global hypomethylation but a moderate trend to global hypermethylation, with repeats responsive to demethylating agents - the repeat-silencing premise is supported by direct data, and the node's claim that it had never been measured is withdrawn
Genericity counter (Roulois 2015, PMID 26317465): viral mimicry - ERV-derived dsRNA, MDA5/MAVS/IRF7 activation - is a class effect of DNA-demethylating agents defined in colorectal cancer, so its occurrence in a KMT2A-r blast would demonstrate a property of the drug, not of the fusion; only the comparative magnitude across methylation subgroups has fusion-relevant content
Shared clinical counter (PMID 38867582): both entries rest on the idea that reversing hypermethylation in t(4;11) infant ALL is therapeutically useful; COG AALL15P1 demethylated these patients successfully and changed nothing about survival, so both must now argue from a biomarker-selected or immunotherapy-paired design rather than from demethylation alone
Counter-evidence and the correlative opportunity: AALL15P1 demethylated infants with KMT2A-r ALL (WGBS, 87% of tested samples) without a survival effect, refuting the chemosensitisation arm of this hypothesis; it also banked serial material in which TE/ERV derepression and an interferon signature could be measured retrospectively and never were
both rest on the same premise, that t(4;11) is a globally hypermethylated genome
Named by this hypothesis as the analysis that tests it; the reference was dangling until now.
TE and ERV loci as a library element class