Every niche-disrupting drug strips the patient's own stem cells of the same shelter — the bite-back that applies to the whole beat
also called on-target niche toxicity · normal HSC mobilisation risk · niche disruption collateral
The bone-marrow hiding place that shelters leukaemia cells is the same hiding place that keeps a child's healthy blood stem cells alive and in position. Any drug that breaks it breaks it for both, at exactly the time the child needs those healthy stem cells most.
The same thing, for a scientist
A single risk entry covering the shared liability of every intervention proposed in this beat. The molecules are not leukaemia-specific: **CXCL12-CXCR4** maintains the normal HSC pool and its blockade is the licensed mechanism of plerixafor for mobilising donor stem cells [E1]; **SCF from endothelium and LEPR+ perivascular cells** is required for normal HSC maintenance [E2]; **NG2** marks both KMT2A-rearranged blasts and the periarteriolar pericytes whose depletion induces HSC cycling and reduces long-term repopulating HSCs [E3]; **ITGA4/VLA-4** anchors normal progenitors as well as blasts. The consequences are specific and foreseeable: loss of the long-term repopulating pool, delayed count recovery, and - in a disease presenting with hyperleukocytosis in most infants - the hazard of **mobilising blasts into the circulation**. Two exceptions are worth naming because they are the shape a safe niche target should have: endothelial ANGPTL2-sEV blockade and P2X7 deletion both reduced leukaemia **with no effect on normal haematopoiesis** in MLL-AF9 models [E4].
The shared liability, molecule by molecule
| intervention | leukaemic target | the same molecule in the normal child |
|---|---|---|
| plerixafor, motixafortide, POL5551 | CXCR4 on blasts | CXCR4 on HSCs; induced deletion collapses the HSC pool 1. Mobilising donor stem cells is the drug's licensed use |
| natalizumab | ITGA4 on KMT2A-r blasts | VLA-4 anchors normal progenitors to VCAM-1 and fibronectin |
| CSPG4/NG2 CAR-T | NG2 on KMT2A-r blasts | NG2 on periarteriolar pericytes; their depletion induces HSC cycling and reduces long-term repopulating HSCs 3 |
| LEPR antagonism | LEPR on blasts | LEPR marks the perivascular stromal cell that is a required source of SCF and CXCL12 2 |
| beta-2 blockade | stromal ADRB2 | adrenergic tone also regulates normal HSC egress, through ADRB3 |
The pattern is that niche targets are shared by construction: what makes a molecule a niche molecule is that it supports primitive haematopoietic cells, and a leukaemic stem cell is a primitive haematopoietic cell.
The specific harms, and the two counterexamples
Harm 1: stem-cell pool depletion. Forcing HSCs out of quiescence, whether by removing NG2+ pericytes, megakaryocytes or CXCL12, expands the cycling pool at the cost of the long-term repopulating pool. In a child who will need that pool for seventy years, and who may need it to recover from further cytotoxic therapy, that is not a transient toxicity.
Harm 2: delayed count recovery. Marrow adipocytes suppress engraftment and fatless mice engraft faster; disrupting perivascular SCF does the opposite. Any manipulation of the niche during or after induction acts on the compartment whose job is to restore the counts.
Harm 3: mobilisation into a hyperleukocytic circulation. Most infants with KMT2A-rearranged ALL present with WBC at or above 50 x 10^9/L (see mechanism-kmt2a-r-tissue-localization and concept-hyperleukocytosis-emergency). Plerixafor mobilised blasts a median 3.4-fold in relapsed children with low disease burden; the same manoeuvre at diagnosis is a different proposition and the safety case would have to precede the efficacy case.
Two counterexamples that define what a safe niche target looks like. In MLL-AF9 models, blocking small-extracellular-vesicle secretion from endothelium delayed leukaemia with no effect on normal haematopoiesis, and P2x7 deletion cut functional leukaemia-initiating cells about 5-fold with no effect on normal haematopoiesis 4. Both identify a mechanism the leukaemia uses and the normal cell does not. That is the property to select for, and it is the property that CXCR4, VLA-4, NG2 and LEPR all lack.
| kind | essential-function |
|---|---|
| severity | high in principle, low in the one paediatric trial that tested it (no dose-limiting toxicity with short-course plerixafor) |
| mechanism | niche molecules are shared by construction: what makes a molecule a niche molecule is that it supports primitive haematopoietic cells, and a leukaemic stem cell is a primitive haematopoietic cell |
| mitigation | select targets the leukaemia uses and the normal cell does not: endothelial ANGPTL2-sEV/LILRB2 and marrow-ATP/P2X7 both reduced MLL-AF9 leukaemia with no effect on normal haematopoiesis. Otherwise keep exposure short and reversible, which is what made plerixafor tolerable in children |
affects
[ "drug-revumenib", "drug-venetoclax" ]
affects untracked agents
[ "plerixafor", "motixafortide", "POL5551", "natalizumab", "CSPG4 CAR-T", "Allo-aca (LEPR antagonist)", "propranolol / beta-2 antagonists" ]
specific harms
[ "depletion of the long-term repopulating HSC pool by forcing quiescent cells into cycle", "delayed count recovery during and after induction", "mobilisation of blasts into a circulation already at risk of hyperleukocytosis and leukostasis in most infants with this disease" ]
counterexamples defining a safe target
[
{
"target": "endothelial small extracellular vesicles / ANGPTL2-LILRB2",
"pmid": "33108353",
"normal_haematopoiesis": "unaffected",
"title": "ANGPTL2-containing small extracellular vesicles from vascular endothelial cells accelerate leukemia progression.",
"authors": "Huang D, Sun G, Hao X, et al.",
"venue": "J Clin Invest",
"year": 2021,
"title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
},
{
"target": "marrow ATP / P2X7",
"pmid": "33301426",
"normal_haematopoiesis": "unaffected",
"title": "Bone marrow niche ATP levels determine leukemia-initiating cell activity via P2X7 in leukemic models.",
"authors": "He X, Wan J, Yang X, et al.",
"venue": "J Clin Invest",
"year": 2021,
"title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
}
]- This entry is itself a bite-back and its own bite-back is that over-weighting it would block a whole class of interventions on theoretical grounds, when the only paediatric trial that tested one found no dose-limiting toxicity.
- Selecting only for leukaemia-selective niche targets restricts the field to two mouse findings, neither of which has a drug.
- Short reversible exposure preserves safety at the cost of the very thing the mechanism needs - sustained eviction from the niche - so the safety mitigation and the efficacy requirement pull in opposite directions.
- How short is short enough? Plerixafor's tolerability in children came from a five-day schedule; no one has established the exposure at which niche blockade begins to cost long-term repopulating capacity.
- Does a CSPG4-directed cell therapy deplete periarteriolar pericytes? Nobody has counted them after such treatment, and it is a measurable question in any existing model.
- Are ANGPTL2-LILRB2 and ATP-P2X7 genuinely leukaemia-selective, or does selectivity disappear over a longer exposure than a mouse experiment measures?
- In an infant presenting with a white count above 300 x 10^9/L, what is the safe way to test a mobilising agent at all, given that mobilisation is the pharmacodynamic endpoint?
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Induced deletion of CXCR4 in adult mice severely reduces HSC numbers and increases sensitivity to myelotoxic injury without impairing expansion of more mature progenitors, and most HSCs contact CXCL12-abundant reticular cells - establishing that the CXCL12-CXCR4 axis targeted by plerixafor in leukaemia is the same axis that maintains the normal quiescent HSC pool.
Maintenance of the hematopoietic stem cell pool by CXCL12-CXCR4 chemokine signaling in bone marrow stromal cell niches · Sugiyama T, Kohara H, Noda M, Nagasawa T · Immunity 2006What this evidence cannot show
Mouse; genetic deletion is more complete than pharmacological antagonism, and clinical plerixafor exposure is brief and reversible, which is precisely why it is tolerated in healthy donors. The risk described here scales with duration and completeness of blockade, and a short course may carry little of it. Read: abstract only.
Evidence that argues against this (1)
- Plerixafor's clinical safety record argues that the risk is manageable in practice: a phase 1 in children with relapsed or refractory acute leukaemia found NO dose-limiting toxicities across four dose levels, with febrile neutropenia and hypokalaemia the commonest attributable grade 3 or higher non-haematological events. Short-course CXCR4 blockade in children is demonstrably tolerable, so on-target niche toxicity is a theoretical liability that a real trial did not encounter. Read: abstract only. Cooper TM et al. 2017
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
HSCs are depleted when Scf is conditionally deleted from endothelial cells or from leptin-receptor-expressing perivascular stromal cells, and most are lost when it is deleted from both - so the LEPR-positive perivascular cell whose leptin-receptor signalling has been proposed as an anti-leukaemic target is also a required source of a normal HSC maintenance factor.
Endothelial and perivascular cells maintain haematopoietic stem cells · Ding L, Saunders TL, Enikolopov G, Morrison SJ · Nature 2012What this evidence cannot show
Mouse conditional genetics. LEPR as a stromal marker and LEPR as a receptor on the blast are different things, and a LEPR antagonist acting on the blast does not necessarily ablate the stromal cell; this evidence establishes the overlap of the marker, not equivalence of the intervention. Read: abstract only.
Evidence that argues against this (1)
- The LEPR antagonist reported in this disease restored chemosensitivity WITHOUT altering baseline leukaemia growth, which is the signature of a selective resistance-reversing agent rather than a broadly cytotoxic one, and nothing in that report describes a haematopoietic toxicity. A receptor-blocking peptide acting on blasts is not equivalent to deleting a stromal population. Read: abstract only. Liao X et al. 2026
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Conditional depletion of NG2-positive periarteriolar pericytes induces HSC cycling and reduces functional long-term repopulating HSCs in the bone marrow, and quiescent HSCs associate specifically with the small arterioles these cells ensheath - so an NG2/CSPG4-directed therapeutic aimed at KMT2A-rearranged blasts would engage a population whose loss depletes the normal long-term stem-cell pool.
Arteriolar niches maintain haematopoietic stem cell quiescence · Kunisaki Y, Bruns I, Scheiermann C, Ahmed J, Pinho S, Zhang D, et al. · Nature 2013What this evidence cannot show
Mouse; diphtheria-toxin depletion of NG2+ cells is far more complete than any antibody or CAR-T would achieve against a perivascular target behind a vessel wall, so this is an upper bound on the hazard. The extrapolation from pericyte depletion to CSPG4 CAR-T toxicity is this entry's inference, not a published observation. Read: abstract only.
Evidence that argues against this (1)
- The concern may be smaller than it looks, because NG2 expression on the leukaemic blast is what the KMT2A literature is about and NG2 is described as minimally expressed in healthy haematopoietic cells - the periarteriolar pericyte is stromal, behind a vessel wall, and a poor target for a cell therapy directed at circulating and marrow-resident blasts. No study has measured pericyte numbers after CSPG4-directed therapy, so the hazard is unmeasured rather than demonstrated. Read: abstract only. Lopez-Millan B et al. 2024
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Two niche-directed manipulations in MLL-AF9 models reduced leukaemia while sparing normal haematopoiesis: reducing small-extracellular-vesicle production from Cdh5+/Tie2+ endothelial cells markedly delayed leukaemia progression and had no effect on normal haematopoiesis, and P2x7 deletion notably impaired leukaemia-initiating cell homing and self-renewal with an approximately 5-fold decrease in functional LICs and no effect on normal haematopoiesis.
Bone marrow niche ATP levels determine leukemia-initiating cell activity via P2X7 in leukemic models · He X, Wan J, Yang X, Zhang X, Huang D, Li X, Zou Y, Chen C, et al. · Journal of Clinical Investigation 2021What this evidence cannot show
Mouse MLL-AF9; 'no effect on normal haematopoiesis' in a short-term mouse experiment is a weaker statement than long-term safety in a child, and P2X7 is broadly expressed on immune cells so systemic effects are not excluded. The endothelial sEV arm comes from a separate paper (PMID 33108353) and is described here alongside. Read: abstract only.
Evidence that argues against this (1)
- Sparing normal haematopoiesis in a mouse over weeks is not the same as sparing a child's stem-cell reserve over decades, and none of the niche-selective manipulations has been tested pharmacologically, let alone in a growing organism. The two counterexamples define a property to aim for, not a demonstrated safe therapy. The one niche-directed agent actually given to children, plerixafor, had no dose-limiting toxicity but also little benefit. Read: abstract only. Cooper TM et al. 2017
the shared on-target liability of every intervention in this beat
CXCR4 blockade mobilises normal HSCs; that is its licensed use
VLA-4 anchors normal progenitors as well as blasts
forcing quiescent HSCs into cycle depletes the long-term repopulating pool
mobilising blasts in a disease that presents with hyperleukocytosis carries a leukostasis hazard
endothelial ANGPTL2-sEV blockade is one of the two manipulations that spared normal haematopoiesis
the licensed mechanism of this drug is the risk this entry describes
unlike CXCR4 antagonism, E-selectin blockade is reported to protect endogenous HSCs