Hypothesis: treatment-remodelled niche, not niche-imposed dormancy, is what residual KMT2A-rearranged disease exploits
also called treatment-remodelled niche hypothesis · niche damage precedes relapse
The usual story is that leukaemia cells hide in a protected corner of the bone marrow and sleep through chemotherapy. The evidence assembled in this beat points somewhere else: the leukaemia and the chemotherapy both damage the marrow's support structures, and it is the damaged, rebuilt marrow — not a pristine hiding place — that the surviving cells grow back in.
The same thing, for a scientist
Standing model: a stable protective niche shelters dormant leukaemic cells, so disrupting it (CXCR4 blockade) plus chemotherapy should clear residual disease. Four observations from this beat fit that model poorly. (i) MLL-AF9 leukaemia **destroys** the quiescence-enforcing NG2+ periarteriolar compartment rather than occupying it, and denervates the marrow. (ii) **Chemotherapy produces the same niche lesions** — sympathetic neuropathy, fatty replacement, altered marrow pO2, megakaryocyte loss with an FGF1-driven expansion switch. (iii) Where niche protection is documented in this genotype it is **metabolic** (CXCL12-dependent redox control, stromal OXPHOS and glutathione transfer, adipocyte leptin driving complex I), not dormancy-based. (iv) **Hypoxia sensitises** KMT2A-rearranged AML to venetoclax plus a menin inhibitor, the opposite of a sanctuary prediction. The hypothesis: **the niche relevant to relapse is the post-treatment niche, and its relevant property is metabolic support rather than imposed dormancy.** If true, the useful intervention is timed to the remodelled marrow and directed at metabolic coupling or adhesion (`mechanism-itga4-niche-adhesion-kmt2a`), not at chemokine-mediated retention. Three tests and one analysis are specified, with falsifiers.
What the standing model predicts and what this beat found instead
| standing model | what the literature in this beat says |
|---|---|
| a stable protective niche exists and leukaemic cells occupy it | MLL-AF9 leukaemia destroys NG2+ periarteriolar cells and disrupts sympathetic nerves; the niche the cells end up in is one they built |
| chemotherapy is neutral with respect to the niche | chemotherapy induces niche sympathetic neuropathy, fatty marrow replacement, megakaryocyte loss with a TGF-beta-to-FGF1 switch, and marked pO2 change |
| protection works by imposing dormancy | in this genotype the documented mechanisms are metabolic: CXCL12-dependent redox control, stromal OXPHOS and glutathione transfer, leptin-driven complex I activity |
| the hypoxic niche protects | hypoxia sensitised KMT2A-rearranged AML to venetoclax plus MI-503 |
| disrupting retention plus chemotherapy clears residual disease | plerixafor mobilised blasts 3.4-fold in children with target engagement proportional to surface CXCR4, and produced three responses in eighteen, none in ALL |
None of these individually refutes the standing model. Together they suggest the model is pointed at the wrong property of the niche and the wrong moment in treatment.
Predictions, and what would kill the hypothesis
Predictions.
- Marrow niche architecture at end of induction differs measurably from diagnosis in the same patient — reduced sympathetic fibre density, reduced NG2+ periarteriolar pericytes, increased adiposity — and the magnitude predicts measurable residual disease or relapse independently of blast burden.
- Stromal co-culture confers little or no protection against a menin inhibitor, in contrast to the protection it confers against cytarabine.
- A CXCR4 antagonist loses efficacy when co-administered with an antioxidant, because its contribution is to a redox axis rather than to retention.
- Venetoclax plus a menin inhibitor clears the deep peri-sinusoidal (most hypoxic) compartment preferentially.
Falsifiers. Any of the following would kill it: stromal co-culture protecting KMT2A-rearranged cells from a menin inhibitor as strongly as from cytarabine; niche architecture at end of induction being indistinguishable from diagnosis; an antioxidant failing to change CXCR4-antagonist efficacy; or hypoxia protecting rather than sensitising in a replication of PMID 38053201 with a clinical menin inhibitor.
| statement | In KMT2A-rearranged leukaemia the niche relevant to residual disease and relapse is the niche as remodelled by the leukaemia and by treatment, and its relevant property is metabolic and adhesive support rather than imposed dormancy; therefore interventions should be directed at metabolic coupling or integrin adhesion in the post-induction marrow rather than at chemokine-mediated retention. |
|---|---|
| origin | literature-gap |
| prior probability | 0.3 |
| what it does not explain | the reversible dormancy of patient-derived ALL xenograft subpopulations on removal from the niche, which is direct evidence for a niche-imposed dormant state in the right disease and is not accounted for here |
predictions
[ "niche architecture at end of induction differs measurably from diagnosis in the same patient and the magnitude predicts MRD or relapse independently of blast burden", "stromal co-culture confers little or no protection against a menin inhibitor, unlike against cytarabine", "a CXCR4 antagonist loses efficacy when co-administered with an antioxidant", "venetoclax plus a menin inhibitor clears the deep peri-sinusoidal compartment preferentially" ]
experiments
[ "exp-cxcr4-antagonist-antioxidant-rescue", "exp-stromal-coculture-menin-inhibitor-protection", "exp-beta2-blockade-kmt2a-xenograft" ]
analyses
[ "ana-marrow-niche-rejected-candidates" ]
falsifiers
[ "stromal co-culture protects KMT2A-rearranged cells from a menin inhibitor as strongly as from cytarabine", "marrow niche architecture at end of induction is indistinguishable from diagnosis in paired human trephines", "an antioxidant does not change CXCR4-antagonist efficacy in a KMT2A-rearranged model", "hypoxia protects rather than sensitises when PMID 38053201 is replicated with a clinical menin inhibitor", "dormant KMT2A-rearranged blasts are shown to reside at NG2+ arterioles and to depend on that residence for survival" ]
- If the hypothesis is right, the therapeutic window is after induction, in a child who is neutropenic, infected and least able to tolerate an additional agent.
- Restoring the niche to its pre-treatment state would restore the compartment that supports normal HSCs and also the compartment that metabolically supports leukaemic cells; the direction of net benefit is unknown.
- The hypothesis makes chemotherapy partly responsible for the conditions of relapse, which is a claim with clinical and ethical weight and must not be stated as established on the strength of a mouse literature and one Letter.
- Are paired diagnosis and end-of-induction trephines available in any infant ALL biobank? The human test of this hypothesis requires paired tissue and nothing else exotic.
- Does the remodelled niche differ functionally, or only structurally? Damage is documented; permissiveness is assumed.
- If metabolic coupling is the mechanism, is it contact-dependent (tunnelling nanotubes, mitochondrial transfer) or soluble? The two imply different drugs.
- Could niche architecture at end of induction be scored from routine diagnostic histology with immunohistochemistry for tyrosine hydroxylase, NG2 and perilipin, rather than requiring research-grade tissue?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
MLL-AF9 AML development disrupts marrow sympathetic nerves and the quiescence of Nestin-positive niche cells, expanding osteoblast-primed mesenchymal progenitors at the expense of HSC-maintaining NG2-positive periarteriolar niche cells, with the leukaemogenic adrenergic signal transduced by the beta-2 and not the beta-3 receptor on stromal cells - so the leukaemia remodels the compartment rather than merely occupying it.
Acute myelogenous leukemia-induced sympathetic neuropathy promotes malignancy in an altered hematopoietic stem cell niche · Hanoun M, Zhang D, Mizoguchi T, Pinho S, Pierce H, Kunisaki Y, Lacombe J, Armstrong SA, et al. · Cell Stem Cell 2014What this evidence cannot show
Mouse, myeloid, MLL-AF9; no lymphoid or infant arm. The remodelling is documented at established disease, not serially through treatment, so 'precedes relapse' is not something this paper tests. Read: abstract only.
Evidence that argues against this (1)
- The dormancy model has direct support in the right disease that this hypothesis must accommodate rather than dismiss: in patient-derived ALL xenografts a rare subpopulation combining long-term dormancy, treatment resistance and stemness matched the expression profile of primary ALL cells recovered at measurable residual disease, and its adverse properties were REVERSIBLE on removal from the in vivo niche. A niche-imposed, reversible dormant state is exactly what that experiment shows, and this hypothesis does not explain it away. Read: abstract only. Ebinger S et al. 2016
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Marrow niche sympathetic neuropathy is present across multiple haematological malignancy models and in human MPN and AML marrow, involves severe ultrastructural fibre damage on electron microscopy, and - per the authors' closing statement - is also induced by chemotherapy, which is first-line treatment for the diseases in which the neuropathy is pathogenic.
Bone marrow sympathetic neuropathy is a hallmark of hematopoietic malignancies and it involves severe ultrastructural damage · Bernal A, Cuminetti V, Serulla M, Florit A, Konieczny J, Golnarnik G, Chen Y, Ferre M, et al. · Experimental Hematology and Oncology 2025What this evidence cannot show
PubMed record is a Letter. The chemotherapy-induces-neuropathy statement is a citation to prior work, not a result of this letter. Neither of the two new models is KMT2A-rearranged. Read: abstract only.
Evidence that argues against this (1)
- A lesion present in every driver tested and also produced by the treatment cannot by itself distinguish a mechanism that residual disease exploits from a non-specific consequence of marrow disease and cytotoxic therapy. The hypothesis needs the remodelled niche to be functionally permissive, not merely damaged, and no study has shown that. Read: abstract only. Zhao M et al. 2014
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Hypoxia induction SENSITISED KMT2A-rearranged AML to venetoclax plus MI-503 for proliferation inhibition and apoptosis induction, with the combination repressing HIF-1A and its target HDAC9, and HDAC9 indispensable for KMT2A-rearranged AML proliferation.
Repressing HIF-1alpha-induced HDAC9 contributes to the synergistic effect of venetoclax and MENIN inhibitor in KMT2Ar AML · Ling Q, Zhou Y, Qian Y, Qian J, Zhang Y, Wang J, Zhu Y, Zhou Y, et al. · Biomarker Research 2023What this evidence cannot show
Letter; cell lines, primary ex vivo cells and a MOLM-13 xenograft; culture-induced hypoxia rather than a measured marrow compartment; MI-503 is a tool compound; myeloid and adult-derived. Unreplicated. Read: abstract only.
Evidence that argues against this (1)
- Three independent niche compartments protect this leukaemia by buffering oxidative stress, which predicts that an oxidative-stress-adapted state should be PROTECTIVE - the opposite of the sensitisation this claim rests on. The hypothesis leans on a single Letter against a larger body of metabolic-protection work, and the two have never been tested with the same drug. Read: abstract only. Forte D et al. 2020
- depends on →MLL-AF9 leukaemia destroys the nerves that run its own niche — a mechanism class this graph had nothing on
built on the finding that the leukaemia destroys rather than occupies the quiescence-enforcing compartment
argues that metabolic support, not imposed dormancy, is the niche property relevant to residual disease; does not account for reversible xenograft dormancy
reframes the documented chemoprotection as metabolic and post-treatment rather than dormancy-based
the megakaryocyte TGF-beta1 to FGF1 switch after chemotherapy is the graph's existing instance of treatment remodelling the niche
prediction 2: stroma confers little or no protection against a menin inhibitor
prediction 3: antioxidant co-treatment rescues the leukaemia from CXCR4 blockade
tests whether preserving adrenergic niche signalling alters leukaemia progression