COG AALL15P1 — azacitidine epigenetic priming in infant KMT2A-rearranged ALL: on-target, well tolerated, no survival effect
also called AALL15P1 · NCT02828358 · azacitidine infant ALL
Infant KMT2A-rearranged leukemia cells have too many chemical tags silencing their DNA, so this trial gave a drug that strips those tags off before each round of chemotherapy. The drug did exactly what it was meant to do to the DNA, was safe, and did not help the babies live longer.
The same thing, for a scientist
AALL15P1 tested five days of azacitidine immediately before chemotherapy in each of four post-induction courses for infants under one year with newly diagnosed KMT2A-rearranged ALL, on the rationale that these blasts are characterised by DNA hypermethylation and that DNA methyltransferase inhibition increases chemotherapy cytotoxicity preclinically. Dose-limiting toxicity occurred in 2 of 31 evaluable patients (6.5%). Whole-genome bisulfite sequencing showed decreased DNA methylation in 87% of peripheral blood mononuclear cell samples tested. Event-free survival was similar to prior studies of newly diagnosed infant ALL, and the authors conclude that incorporating azacitidine to enhance cytotoxicity did not impact survival.
A clean experiment with a clean negative1 deeper
This trial deserves attention out of proportion to its result, because it is methodologically exemplary in a way most negative trials are not.
- The biological premise was specific and correct: infant KMT2A-rearranged ALL blasts are DNA-hypermethylated. This graph covers the methylation phenotype under
mechanism-kmt2a-r-dna-methylation-signature. - The pharmacodynamic endpoint was measured and met: whole-genome bisulfite sequencing showed decreased DNA methylation in 87% of samples tested after five days of azacitidine 1. The drug did what it was supposed to do, in patients, and it was proved.
- The safety endpoint was met: 2 of 31 evaluable patients (6.5%) had dose-limiting toxicity. In infants, on top of four intensive chemotherapy courses, that is genuinely well tolerated.
- And survival did not change.
That combination — correct premise, confirmed target engagement, acceptable toxicity, no benefit — is the most informative kind of negative trial, because it eliminates the usual excuses. It was not underdosed, it was not too toxic, and the mechanism was not wrong about what it does. Hypomethylation simply is not sufficient to make these blasts more chemo-sensitive in vivo.
The tension with the preclinical literature, and its likely resolution
Preclinical work in infant KMT2A-rearranged ALL showed that hypomethylating agents are additive with most conventional chemotherapeutic agents and that azacitidine and decitabine prolong survival in xenografts. But the same study found that in a subset of samples hypomethylating agents were antagonistic with several agents, most evidently methotrexate, apparently through upregulation of ABC transporters 2.
Methotrexate is a backbone drug in Interfant protocols. If azacitidine antagonises methotrexate in some patients while sensitising to other agents in others, a trial that gives azacitidine before a multi-agent course could easily average to nothing — which is what happened. The same preclinical work found that the combination with a clear in vivo survival advantage was azacitidine plus venetoclax, not azacitidine plus chemotherapy.
So the actionable reading of AALL15P1 is not 'hypomethylation does not work in infant ALL'. It is 'hypomethylation plus chemotherapy does not work, and the partner the preclinical data actually supported was venetoclax' — which is now being randomised in infants in NCT06317662, and paired with menin inhibitors in the paediatric trials NCT06177067 and NCT06397027.
| nct | NCT02828358 |
|---|---|
| phase | 2 (groupwide pilot) |
| status | Completed |
| open now | false |
| pediatric eligible | true |
| population | Infants <1 year with newly diagnosed KMT2A-rearranged ALL |
| n registered | 78 |
| n evaluable | 31 |
| intervention | azacitidine for 5 days immediately before chemotherapy starting on day 6, in each of four post-induction courses |
| efficacy result | event-free survival similar to prior studies; azacitidine did not impact survival (no numeric EFS given in the primary abstract) |
| readout | Negative for survival despite confirmed on-target pharmacodynamics and acceptable toxicity |
primary result
{
"dlt_n": 2,
"dlt_denominator": 31,
"dlt_pct": 6.5,
"verdict": "safe and well tolerated"
}pharmacodynamic result
{
"samples_with_decreased_methylation_pct": 87,
"assay": "whole genome bisulfite sequencing of PBMCs"
}- The most sobering lesson in this beat: confirmed target engagement in patients, with acceptable toxicity, and no survival benefit. Pharmacodynamic success is not efficacy, and this trial proves it cleanly in exactly the population and genotype of interest.
- Combination antagonism: a rationally chosen agent can make a backbone drug work worse, and averaging across a multi-agent protocol hides it. Every new combination in this disease should be checked against the methotrexate antagonism finding.
- Epigenetic therapy in infancy: azacitidine given during the period of maximal developmental DNA methylation remodelling has unknown long-term consequences, and a pilot with 31 patients and short follow-up cannot see them.
- Underpowering as a hazard: a 31-patient pilot reporting 'no impact on survival' will be cited as evidence that hypomethylation does not work, when in fact it could not have detected a clinically meaningful benefit.
- Was azacitidine antagonising methotrexate in these infants? The preclinical ABC-transporter antagonism finding offers a specific, testable explanation for a null result in a multi-agent protocol, and the trial's banked samples could be interrogated for transporter expression.
- Would azacitidine plus venetoclax — the combination the preclinical data actually supported — work where azacitidine plus chemotherapy did not? NCT06317662 randomises venetoclax in infants but on a blinatumomab rather than an azacitidine backbone, so the specific question remains open.
- Does azacitidine priming change menin-inhibitor sensitivity? Hypomethylation and menin eviction act on the same silenced loci from different directions, and three paediatric trials are already giving azacitidine with a menin inhibitor without anyone having tested the interaction directly.
- Should PBMC methylation be accepted as a pharmacodynamic surrogate for blast methylation? The trial's key on-target evidence rests on it and the two compartments are not the same.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
COG AALL15P1 gave five days of azacitidine immediately before chemotherapy in four post-induction courses to infants under one year with newly diagnosed KMT2A-rearranged ALL; dose-limiting toxicity occurred in 2 of 31 evaluable patients (6.5%), whole-genome bisulfite sequencing showed decreased DNA methylation in 87% of peripheral blood mononuclear cell samples tested, event-free survival was similar to prior studies of newly diagnosed infant ALL, and incorporating azacitidine to enhance cytotoxicity 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. · Haematologica 2024What this evidence cannot show
A single-arm pilot with 31 evaluable patients and a safety/tolerability primary endpoint; it was never powered to detect a survival difference, so 'did not impact survival' means 'no signal in a study that could not have detected a modest one'. No numeric EFS percentage is given in the abstract and none is asserted here — the comparison to 'prior studies' is descriptive. Methylation was measured in peripheral blood mononuclear cells rather than in marrow blasts specifically.
Evidence that argues against this (1)
- The null result may be a schedule artefact rather than a verdict on hypomethylation. Preclinical work in KMT2A-rearranged infant ALL found hypomethylating agents to be ANTAGONISTIC with methotrexate in a subset of samples (attributed to ABC transporter upregulation) while azacitidine plus venetoclax gave a significant in vivo survival advantage. AALL15P1 gave azacitidine as priming immediately before conventional cytotoxic courses that include methotrexate, which is the one pairing the preclinical data argue against; the same agent in the pairing the preclinical data support has never been tested in infants. Cheung LC et al. 2023
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
In preclinical models of infant KMT2A-rearranged ALL, hypomethylating agents were additive with most conventional chemotherapeutic agents but antagonistic in a subset of samples, most evidently with methotrexate via apparent ABC transporter upregulation, while azacitidine plus venetoclax produced a significant in vivo survival advantage.
Preclinical efficacy of azacitidine and venetoclax for infant KMT2A-rearranged acute lymphoblastic leukemia reveals a new therapeutic strategy · Cheung LC, Aya-Bonilla C, Cruickshank MN, et al. · Leukemia 2022What this evidence cannot show
Cell lines and xenografts; the antagonism finding is in a subset of samples and the mechanism (ABC transporter upregulation) is inferred. This study postdates the design of AALL15P1, so it is an explanation offered in hindsight rather than a prediction the trial ignored.
Evidence that argues against this (1)
- The clinical result of this preclinical rationale is now published and it is null for outcome. AALL15P1 delivered five days of azacitidine immediately before chemotherapy in four post-induction courses to infants with newly diagnosed KMT2A-rearranged ALL: dose-limiting toxicity in 2 of 31 evaluable patients (6.5%), decreased DNA methylation on whole-genome bisulfite sequencing of peripheral blood mononuclear cells in 87% of tested samples — and EVENT-FREE SURVIVAL SIMILAR TO PRIOR STUDIES of newly diagnosed infant ALL. The preclinical antagonism with methotrexate is therefore not the reason the strategy did not deliver; the strategy did not deliver even where the epigenetic mechanism was demonstrably engaged. Guest EM et al. 2024
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.trial_registry
AALL15P1 is registered as NCT02828358, a completed phase 2 NCI-sponsored groupwide pilot study with registered enrolment of 78, started April 2017, testing the tolerability and biologic activity of adding azacitidine to chemotherapy in infants with ALL and KMT2A gene rearrangement.
A Groupwide Pilot Study to Test the Tolerability and Biologic Activity of the Addition of Azacitidine to Chemotherapy in Infants With Acute Lymphoblastic Leukemia and KMT2A Gene Rearrangement · National Cancer Institute (NCI); Children's Oncology Group · ClinicalTrials.gov 2024What this evidence cannot show
Registered enrolment (78) exceeds the 31 evaluable patients reported, reflecting enrolment versus evaluability rather than a discrepancy.
Evidence that argues against this (1)
- The registered enrolment of 78 and the number the trial's primary endpoint actually rests on are different numbers, and the registered outcome is tolerability, not efficacy. Raw ClinicalTrials.gov v2 API, retrieved 2026-09-19: the SOLE primary outcome of NCT02828358 is 'Tolerability of Azacitidine in Combination With Interfant-06 Standard Chemotherapy in Evaluable Infant Patients' over 6 months; ACTUAL enrolment is 78, primary completion 2020-09-30, and results have been posted since 2021-10-04. The peer-reviewed report of the same trial states that the tolerability endpoint rested on 31 EVALUABLE patients (2 with dose-limiting toxicity, 6.5%). A pilot powered for tolerability in 31 evaluable infants cannot support or refute an efficacy claim about azacitidine, and the registered enrolment figure conceals that. Guest EM et al. 2024
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
Counter-evidence: the demethylation therapy this miRNA-silencing literature motivates was tested unselected in infant KMT2A-r ALL, reached its pharmacodynamic target, and did not change event-free survival - so a further hypomethylating-agent proposal needs a selection biomarker or a different combination partner
The intervention under test: 5 days of azacitidine before chemotherapy in four post-induction courses, NCT02828358, infants under one year.
The clinical expression of the risk-benefit: safe, pharmacodynamically active, no survival effect.
Counter-evidence (Guest 2024, PMID 38867582): AALL15P1 demonstrated the epigenetic mechanism -- decreased DNA methylation in 87% of sampled PBMC after five days of azacitidine, DLT in 2/31 evaluable -- and event-free survival was similar to prior infant ALL studies; target engagement without outcome benefit
azacitidine priming; on-target demethylation in 87% of samples, no survival effect
azacitidine before each of four post-induction courses; safe, DLT 6.5%, no survival impact
the preclinical partner that actually showed in vivo survival benefit was venetoclax, not chemotherapy