Where KMT2A-rearranged leukemia physically sits — marrow niche, CNS, blood, liver/spleen and skin
also called tissue localization of KMT2A-r leukemia · extramedullary disease in infant ALL · CNS involvement infant ALL · bone marrow niche KMT2A-r · hyperleukocytosis infant ALL · leukemia cutis infant · CXCR4/CXCL12 axis in MLL-rearranged ALL
KMT2A-rearranged leukemia is not just in the bone marrow: it spills into the blood in enormous numbers, reaches the fluid around the brain in roughly half of infants, swells the liver and spleen, and in newborns can show up as lumps in the skin. Where the cells sit changes both how they behave and how hard they are to kill, because the marrow and the brain each shelter them in different ways.
The same thing, for a scientist
KMT2A-rearranged (KMT2A-r) acute leukemia has a distinctive anatomical footprint. In the marrow, KMT2A-r ALL blasts are differentially protected by mesenchymal stroma relative to non-KMT2A-r ALL, and both the protection and the leukemia-stem-cell fraction's drug resistance are reversible by CXCR4 antagonism, making the CXCR4/CXCL12(SDF-1) axis the best-characterised niche dependency in this genotype. CNS involvement at diagnosis is far more frequent than in older-child B-ALL: 44.0% of KMT2A-r infants versus 15.4% of KMT2A-germline infants in one cohort (n=139), and only 45.8% of KMT2A-r infants were CNS1 (no CSF blasts) in the COG P9407 trial (n=155 KMT2A-r) — and it is prognostically decisive. Hyperleukocytosis is the rule rather than the exception (WBC ≥50x10^9/L in 79.4% of KMT2A-r infants, 92.2% of KMT2A::AFF1), and WBC is an independent prognostic factor in both Interfant-99 and Interfant-06, where WBC ≥300x10^9/L is a formal high-risk criterion. Hepatosplenomegaly is near-universal in paediatric BCP-ALL (80%), and cutaneous extramedullary disease is a characteristic neonatal presentation in which KMT2A rearrangement is the single commonest lesion (26/50, 52%). Murine work shows that the CNS niche does not merely harbour KMT2A-AFF1 cells but metabolically reprograms them.
The bone marrow niche: a genotype-specific dependency, not a generic one1 deeper
The important finding is comparative. Using primary infant MLL-rearranged (KMT2A-r) ALL samples, MLL-R ALL was differentially protected by stroma from spontaneous apoptosis compared with non-MLL-R ALL — i.e. the niche dependency is not a property of B-ALL in general but is enriched in this genotype. Co-culture of bulk MLL-R ALL either in direct contact with stroma or with stroma-produced soluble factors alone promoted proliferation and cell-cycle entry, and stroma protected both bulk blasts and the leukemia stem cell (LSC) fraction from the cytotoxicity of the FLT3 inhibitor lestaurtinib. In a xenograft model, CXCR4 inhibition with AMD3100 (plerixafor) markedly enhanced lestaurtinib efficacy 1.
That the soluble fraction alone works is mechanistically important: it means the protection is at least partly chemokine/cytokine-mediated rather than requiring adhesion, which is what makes the CXCR4/CXCL12 (SDF-1) axis a plausible pharmacological handle.
The follow-up work dissected the axis. The protein-epitope-mimetic CXCR4 antagonist POL5551 showed nanomolar activity in pre-B and T ALL cell lines and paediatric ALL primary samples: it decreased CXCR4 antibody binding, blocked SDF-1α-mediated ERK1/2 phosphorylation, inhibited SDF-1α-induced chemotaxis, and reversed stromal-mediated protection from chemotherapy. It was significantly more effective at inhibiting CXCR4 antibody binding than plerixafor. Both POL5551 in vitro and cytarabine with or without POL5551 in vivo modulated the surface expression of adhesion molecules, and POL5551 increased cytarabine sensitivity in a xenograft model of infant MLL-rearranged ALL 2.
So the niche dependency has three limbs that have each been demonstrated: chemotaxis/retention (CXCR4-SDF-1), survival signalling (ERK1/2), and adhesion-molecule modulation.
What is NOT established about the marrow niche
I could not verify any of the following, and none is asserted in this node: that KMT2A-r blasts express higher surface CXCR4 than other B-ALL subtypes; that a specific mesenchymal stromal cell subset is required; that CXCR4 antagonism alone (without a partner drug) has activity in KMT2A-r ALL; or that any CXCR4 antagonist has been tested clinically in KMT2A-r infant ALL. The published claim is narrower and should be quoted narrowly: stroma protects KMT2A-r ALL more than non-KMT2A-r ALL, and CXCR4 blockade re-sensitises it to a partner agent in xenografts 12.
Central nervous system: not an occasional complication, closer to a default1 deeper
MLL-Baby cohort (n = 139 consecutive infants with BCP-ALL). CSF was examined by microscopy of cytospin slides. CNS involvement at any level was found in 50 of 139 infants (36.0%), and the incidence was significantly higher with KMT2A rearrangement: 44.0% for KMT2A-r versus 15.4% for KMT2A-germline (P = .003). The prognostic impact was confined to the KMT2A-r group, where event-free survival was 0.21 for CNS-positive versus 0.48 for CNS-negative infants (P = .044). CNS-positive infants could not be salvaged by conventional chemotherapy alone irrespective of how fast their MRD cleared, whereas the combination of initial CNS negativity plus fusion-gene-transcript MRD negativity identified up to one-third of KMT2A-r infants who achieved disease-free survival above 95% on chemotherapy 3.
COG P9407 (n = 209 analysed; 157 (75.1%) KMT2A-rearranged). Using the CNS1/CNS2/CNS3 staging system, CNS1 (no blasts in CSF) occurred in only 71 of 155 (45.8%) KMT2A-r infants versus 32 of 52 (61.5%) KMT2A-germline infants — i.e. more than half of KMT2A-r infants had detectable CSF blasts at diagnosis. CNS status was prognostic: 5-year EFS was 54 ± 6%, 41 ± 8% and 33 ± 10% for CNS1, CNS2 and CNS3 in the treatment-eligible cohort overall (P = 0.025), 48 ± 7%, 27 ± 8% and 30 ± 9% within KMT2A-r (P = 0.0155), and 53 ± 12%, 18 ± 10% and 23 ± 12% in the KMT2A::AFF1 subgroup specifically (P = 0.022) 4.
Two cohorts, two staging methods, one conclusion: CNS disease in KMT2A-r infant ALL is common enough that its absence, rather than its presence, is the informative finding — and it is the difference between a 48-54% and a 27-33% five-year EFS.
The CNS niche actively reprograms the cells it shelters
In a fully murine, immunocompetent KMT2A-AFF1+ mixed-lineage infant leukemia model, previously defined bone-marrow-derived leukemia propagating cells (LPCs) engrafted the CNS, but not equally: lineage-negative cKit+ common lymphoid progenitor cells were consistently underrepresented in the CNS niche. Transplanting CNS-derived LPCs (modelling relapse) produced reduced systemic repopulation capacity, with many recipients showing stable long-term engraftment without developing overt leukemia — a phenomenon not seen with bone-marrow-derived transplants.
Transcriptomic profiling of matched CNS- and BM-derived LPCs from the same animals showed CNS enrichment of hypoxia, lipid and cholesterol homeostasis, and inflammatory signalling pathways; the LPC subsets that successfully adapted to the CNS upregulated lipid and fatty-acid metabolic programs; and CNS-derived LPCs showed increased expression of genes involved in T-cell immune modulation 9.
The interpretation offered — that the CNS imposes selective pressures causing lasting metabolic and functional reprogramming that impairs the cells' ability to re-establish systemic disease — is the opposite of the usual 'sanctuary site' framing. If it generalises, CNS-resident KMT2A-r cells are not simply hiding; they are a metabolically distinct, more immunosuppressive and less systemically aggressive population, which has direct consequences for how CNS relapse should be treated and for whether T-cell-engaging agents can work there.
Blood: hyperleukocytosis is the presenting physiology
How often. In COG P9407, WBC ≥50 x 10⁹/L was present in 123 of 155 (79.4%) KMT2A-r infants, and the fraction was partner-dependent: 71 of 77 (92.2%) for KMT2A::AFF1 versus 6 of 14 (42.9%) for KMT2A::MLLT3, with WBC count differing significantly across AFF1, MLLT1, MLLT3, 'other' and KMT2A-germline groups (P < 0.0001) 4. This is one of the clearest examples in the whole KMT2A literature of the 3' partner setting the clinical phenotype.
How much it matters. In Interfant-99 (482 infants treated on the hybrid protocol; 4-year EFS 47.0%, SE 2.6), all types of MLL (KMT2A) rearrangement, very high white blood cell count, age younger than 6 months, and poor prednisone-prophase response were each independently associated with inferior outcome 6. In Interfant-06 (651 infants; 6-year EFS 46.1%, SE 2.1 and OS 58.2%, SE 2.0), WBC count is a formal risk-stratification variable: the high-risk group was defined as KMT2A-rearranged and older than 6 months and either WBC ≥300 x 10⁹/L or a poor prednisone response. That high-risk group had a 6-year EFS of 20.9% (SE 3.4) even with the intention to transplant — and only 46% actually reached transplant because events came too early. On multivariable analysis, KMT2A rearrangement was the strongest prognostic factor for EFS, followed by age, WBC count, and prednisone response 5.
So the thresholds in use are: ≥50 x 10⁹/L as the descriptive marker of hyperleukocytosis in infant cohorts, and ≥300 x 10⁹/L as the trial-operational high-risk cut-off in Interfant-06.
Liver and spleen
Organomegaly is the physical correlate of the high circulating burden. In a cohort of 364 paediatric precursor-B malignancies, hepatosplenomegaly was present in 80% of BCP-ALL patients (n = 154) versus 2% of B-lymphoblastic lymphoma patients (n = 210), and hepatosplenomegaly together with peripheral blood involvement (95% vs 2%) were the two most important discriminators between leukemic and lymphomatous presentation of the same underlying malignant cell 8.
Scope caution: that 80% is for paediatric BCP-ALL generally, not specifically for KMT2A-r or infant disease. I could not verify a KMT2A-r-specific or infant-specific hepatosplenomegaly percentage from a primary source, so none is stated here. Given that KMT2A-r infants have higher WBC counts than other B-ALL patients 4, 80% is a floor rather than an estimate for this subgroup — but that is an inference, not a measurement.
Skin: leukemia cutis and cutaneous extramedullary disease in neonates2 deeper
Cutaneous involvement is a genuinely characteristic presentation at the youngest ages. In the largest cohort to date — 50 infants diagnosed between 0 and 6 months of life with either isolated cutaneous extramedullary disease (cEMD, n = 8) or cutaneous disease with marrow involvement (AML-cEMD, n = 42), born 1990-2018 across Italy, the Netherlands, Switzerland and the United States — KMT2A rearrangement was the most common genetic lesion, in 26 of 50 (52%). Overall 5-year event-free survival was 66% (95% CI 51-78) and overall survival 75% (95% CI 60-85). CNS involvement was found in 25% of cEMD patients. Complete spontaneous remission occurred without any therapy in two patients. No difference in outcome was observed between the AML-cEMD and cEMD groups, none of the cEMD patients died, and — notably — KMT2A rearrangements were not associated with poorer prognosis in this cohort 7.
The same report frames the background rates: myeloid neoplasms account for 50% of infant leukemias; approximately 25-50% of patients with newborn leukemia have cutaneous extramedullary disease; and in fewer than 10% of patients aleukemic leukemia cutis or isolated cutaneous disease occurs, with skin lesions appearing before marrow involvement and systemic symptoms 7.
Two things in that paragraph should unsettle a reader of the rest of this node. Spontaneous remission of a KMT2A-r-enriched leukemia, and KMT2A rearrangement not carrying a poor prognosis, are both the opposite of the marrow-based KMT2A-r story. Whether that reflects a genuinely different biology of skin-presenting neonatal myeloid neoplasia, or simply a 50-patient cohort spanning 28 years, is unresolved.
Other extramedullary sites
Extramedullary disease outside the CNS is uncommon but not negligible, and it can be the first manifestation of a lineage switch. In the Project EVOLVE post-immunotherapy lineage-switch cohort, non-CNS extramedullary disease was present in 5 of 70 (7.1%) and CNS disease in 10 of 70 (14.3%) at the time of the most proximal immunotherapy 10. A documented case of KMT2A::AFF1 B-ALL on the fourth cycle of blinatumomab presented with back pain and lower-limb weakness from an extradural thoracic mass causing spinal cord compression, which proved on biopsy to be a mixed-phenotype myeloid sarcoma carrying the same KMT2A rearrangement, with concurrent MPAL in the marrow 11. New focal or neurological symptoms in a KMT2A-r patient on antigen-directed therapy therefore warrant imaging and biopsy, not a marrow aspirate alone.
The anatomical footprint shrinks across the first year of life
A multi-institutional review of 54 cases of infantile B-ALL across four medical centres, divided into congenital and non-congenital groups with multiple age subgroups, found decreasing trends of hyperleukocytosis, central nervous system involvement, KMT2A rearrangement, lineage switch and mortality, against increasing trends of CD10 expression and non-KMT2A abnormalities, with statistically significant breakpoints at 3 months and at 9 months — the latter not previously described. By the last trimester of the first year, infantile B-ALL essentially merges with paediatric B-ALL 12.
This is the single best argument that the anatomical distribution described in this node is not 'infant ALL biology' but KMT2A-r biology that happens to be concentrated in infants — the same variables that fall with age are the ones that track KMT2A rearrangement.
| mechanism class | anatomical and microenvironmental localisation |
|---|---|
| age gradient | Hyperleukocytosis, CNS involvement, KMT2A rearrangement, lineage switch and mortality all decline across the first year of life with significant breakpoints at 3 and 9 months; the anatomical footprint described here tracks KMT2A rearrangement rather than infancy per se. |
sites
[
{
"site": "bone marrow",
"niche_dependency": "CXCR4/CXCL12 (SDF-1) axis; mesenchymal stroma contact and soluble factors; ERK1/2 survival signalling; adhesion-molecule modulation",
"genotype_specificity": "KMT2A-r ALL differentially protected by stroma from spontaneous apoptosis versus non-KMT2A-r ALL",
"pharmacological_handles": [
"AMD3100 / plerixafor",
"POL5551"
],
"source_evidence": [
"E1",
"E2"
]
},
{
"site": "central nervous system / CSF",
"frequency_at_diagnosis": [
{
"metric": "CNS involvement at any level (CSF cytospin microscopy)",
"value": 36,
"unit": "%",
"n": 139,
"population": "infant BCP-ALL, MLL-Baby protocol",
"source_evidence": "E3"
},
{
"metric": "CNS involvement at any level, KMT2A-rearranged",
"value": 44,
"unit": "%",
"population": "infant BCP-ALL",
"source_evidence": "E3"
},
{
"metric": "CNS involvement at any level, KMT2A-germline",
"value": 15.4,
"unit": "%",
"population": "infant BCP-ALL",
"source_evidence": "E3"
},
{
"metric": "CNS1 (no CSF blasts), KMT2A-rearranged",
"value": 45.8,
"unit": "%",
"n": 155,
"population": "infant ALL, COG P9407",
"source_evidence": "E4"
},
{
"metric": "CNS1 (no CSF blasts), KMT2A-germline",
"value": 61.5,
"unit": "%",
"n": 52,
"population": "infant ALL, COG P9407",
"source_evidence": "E4"
}
],
"prognostic_weight": "EFS 0.21 (CNS-positive) vs 0.48 (CNS-negative) in KMT2A-r infants, P=.044; 5-year EFS 48+/-7% / 27+/-8% / 30+/-9% for CNS1/CNS2/CNS3 in KMT2A-r (P=0.0155) and 53+/-12% / 18+/-10% / 23+/-12% in KMT2A::AFF1 (P=0.022)",
"niche_biology": "CNS niche imposes hypoxia, lipid/cholesterol and inflammatory signalling; CNS-adapted KMT2A-AFF1 cells upregulate lipid and fatty-acid metabolism and T-cell immune-modulation genes, and lose systemic repopulation capacity (murine)",
"source_evidence": [
"E3",
"E4",
"E9"
]
},
{
"site": "peripheral blood (hyperleukocytosis)",
"thresholds": [
{
"threshold": "WBC >= 50 x 10^9/L",
"role": "descriptive marker of hyperleukocytosis in infant ALL cohorts"
},
{
"threshold": "WBC >= 300 x 10^9/L",
"role": "Interfant-06 high-risk criterion (with KMT2A-r and age >6 months, or poor prednisone response)"
}
],
"frequency": [
{
"metric": "WBC >= 50 x 10^9/L, KMT2A-rearranged",
"value": 79.4,
"unit": "%",
"n": 155,
"source_evidence": "E4"
},
{
"metric": "WBC >= 50 x 10^9/L, KMT2A::AFF1",
"value": 92.2,
"unit": "%",
"n": 77,
"source_evidence": "E4"
},
{
"metric": "WBC >= 50 x 10^9/L, KMT2A::MLLT3",
"value": 42.9,
"unit": "%",
"n": 14,
"source_evidence": "E4"
}
],
"prognostic_weight": "Independent prognostic factor in Interfant-99 ('very high WBC') and the third-strongest factor for EFS in Interfant-06 after KMT2A rearrangement and age; Interfant-06 high-risk group 6-year EFS 20.9% (SE 3.4)",
"source_evidence": [
"E4",
"E5",
"E6"
]
},
{
"site": "liver and spleen",
"frequency": [
{
"metric": "hepatosplenomegaly",
"value": 80,
"unit": "%",
"n": 154,
"population": "paediatric B-cell precursor ALL (not KMT2A-r-specific)",
"source_evidence": "E8"
},
{
"metric": "peripheral blood involvement",
"value": 95,
"unit": "%",
"n": 154,
"population": "paediatric B-cell precursor ALL",
"source_evidence": "E8"
}
],
"note": "No KMT2A-r-specific or infant-specific hepatosplenomegaly percentage could be verified from a primary source; the 80% figure is paediatric BCP-ALL overall.",
"source_evidence": [
"E8"
]
},
{
"site": "skin (leukemia cutis / cutaneous extramedullary disease)",
"frequency": [
{
"metric": "KMT2A rearrangement among infants 0-6 months with cutaneous myeloid neoplasm",
"value": 52,
"unit": "%",
"n": 50,
"source_evidence": "E7"
},
{
"metric": "cutaneous extramedullary disease among newborn leukemia (background rate cited by authors)",
"value": "25-50",
"unit": "%",
"source_evidence": "E7"
},
{
"metric": "aleukemic leukemia cutis / isolated cutaneous EMD preceding marrow involvement",
"value": "<10",
"unit": "%",
"source_evidence": "E7"
},
{
"metric": "CNS involvement among isolated cutaneous EMD patients",
"value": 25,
"unit": "%",
"n": 8,
"source_evidence": "E7"
}
],
"outcome": "5-year EFS 66% (95% CI 51-78), OS 75% (95% CI 60-85); two complete spontaneous remissions without therapy; KMT2A rearrangement NOT associated with poorer prognosis in this cohort",
"source_evidence": [
"E7"
]
},
{
"site": "other extramedullary (myeloid sarcoma, epidural, orbit)",
"frequency": [
{
"metric": "non-CNS extramedullary disease at time of proximal immunotherapy in post-immunotherapy lineage-switch cohort",
"value": 7.1,
"unit": "%",
"n": 70,
"source_evidence": "E10"
}
],
"note": "Extramedullary myeloid sarcoma can be the presenting manifestation of a lineage switch, including epidural mass with spinal cord compression.",
"source_evidence": [
"E10",
"E11"
]
}
]cohorts used — the numbers behind the chart
[
{
"cohort": "MLL-Baby",
"n": 139,
"population": "infant BCP-ALL",
"source_evidence": "E3"
},
{
"cohort": "COG P9407",
"n": 209,
"n_kmt2a_r": 157,
"population": "infant ALL",
"source_evidence": "E4"
},
{
"cohort": "Interfant-06",
"n": 651,
"population": "infant ALL",
"source_evidence": "E5"
},
{
"cohort": "Interfant-99",
"n": 482,
"population": "infant ALL",
"source_evidence": "E6"
},
{
"cohort": "cutaneous infant myeloid neoplasm multicentre cohort",
"n": 50,
"n_kmt2a_r": 26,
"source_evidence": "E7"
}
]- The marrow niche is a drug-resistance reservoir that is specifically stronger in this genotype: stroma protects both bulk KMT2A-r blasts and the leukemia stem cell fraction, so a systemically active agent can look effective in vitro and fail against the cells that matter.
- CXCR4 antagonism as a chemosensitiser mobilises normal haematopoietic stem and progenitor cells too — that is what plerixafor is licensed to do — so the same manoeuvre that flushes leukemia out of the niche also strips the marrow of the progenitors an infant needs to survive the cytotoxic partner drug.
- The CNS is both a sanctuary and, on the murine data, a reprogramming chamber: cells that adapt there acquire lipid-metabolic and immunomodulatory programs and upregulate T-cell-modulating genes, which would blunt any T-cell-engaging therapy exactly where the blood-brain barrier already limits drug exposure.
- Hyperleukocytosis is an immediate mechanical and metabolic hazard independent of the leukemia's drug sensitivity: leukostasis, tumour lysis and coagulopathy can kill before any targeted agent has time to work, and a KMT2A::AFF1 infant arrives hyperleukocytic more than nine times out of ten.
- Extramedullary sites do not report to the marrow aspirate. A patient can be in flow-cytometric marrow remission while a KMT2A-r myeloid sarcoma grows in the epidural space, so a response assessment based on marrow alone will overstate disease control at exactly the moment a lineage switch is emerging.
- Is CSF flow cytometry, rather than cytospin morphology, the right way to stage the CNS in KMT2A-rearranged infants? Cytospin found CNS involvement in 36% of infants and only 45.8% of KMT2A-r infants were CNS1 in COG P9407 — both using morphology. Flow cytometry of CSF detects disease below the morphological threshold, and if the true rate is materially higher, the 'CNS-negative plus MRD-negative' good-risk group with >95% DFS may be defined by an insensitive assay.
- Do KMT2A-rearranged blasts actually express more surface CXCR4 than other B-ALL subtypes, or is the enhanced stromal protection mediated by something else? The published work shows the protection is greater and that CXCR4 blockade reverses it, but a direct quantitative comparison of CXCR4 surface density across B-ALL genotypes in primary infant samples is, as far as I could verify, not published.
- Does the murine finding that CNS-adapted KMT2A-AFF1 cells lose systemic repopulation capacity hold in human disease? If CNS-resident cells are metabolically distinct and less systemically aggressive, then CNS-directed therapy and systemic therapy are addressing biologically different populations, and lipid/fatty-acid metabolism becomes a CNS-specific target. Nobody has profiled matched CSF and marrow blasts from the same KMT2A-r infant.
- Why does KMT2A::AFF1 present with hyperleukocytosis in 92.2% of infants while KMT2A::MLLT3 does so in only 42.9%? This is one of the largest partner-specific clinical differences on record and there is no mechanistic explanation. Candidate answers — different cell of origin, different proliferative output, different marrow retention via adhesion/CXCR4 — are all testable and untested.
- Why is KMT2A rearrangement not prognostically adverse in neonates presenting with cutaneous disease, and why do some of those leukemias remit spontaneously? A 50-patient cohort with 52% KMT2A-r, 75% 5-year OS and two spontaneous remissions describes a KMT2A-r leukemia behaving unlike any other. If that is real rather than a small-cohort artefact, something about the neonatal skin or the neonatal immune system is controlling this disease, and it is the only setting where a KMT2A-r leukemia is known to regress on its own.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Primary infant MLL-rearranged (KMT2A-rearranged) ALL is differentially protected by bone marrow stroma from spontaneous apoptosis compared with non-MLL-rearranged ALL; co-culture with stroma or with stroma-produced soluble factors alone promotes proliferation and cell-cycle entry; stroma protects both bulk blasts and the leukemia stem cell fraction from lestaurtinib cytotoxicity; and in a xenograft model CXCR4 inhibition with AMD3100 (plerixafor) markedly enhanced lestaurtinib efficacy.
MLL-rearranged acute lymphoblastic leukaemia stem cell interactions with bone marrow stroma promote survival and therapeutic resistance that can be overcome with CXCR4 antagonism · Sison EA, Rau RE, McIntyre E, Li L, Small D, Brown P · British Journal of Haematology 2013What this evidence cannot show
In vitro co-culture on a stromal cell line plus xenografts in immunodeficient mice; the number of primary infant MLL-R samples is small (this subtype is rare) and is not stated in the abstract, so no n is asserted. The drug partner is lestaurtinib, a FLT3 inhibitor that failed clinically, so the combination result does not transfer to current therapy. Xenograft niches are murine and reconstituted, which is an imperfect model of the human infant marrow. 'Differentially protected' is a relative statement between two primary-sample groups, not a quantified effect size I could verify.
Evidence that argues against this (1)
- Stromal adhesion is not the whole account of marrow retention in this disease, and the pharmacology says so. A CXCR4 antagonist enhances chemotherapy sensitivity in paediatric ALL, which means the retention that matters is at least partly chemokine-driven rather than adhesion-driven. More awkwardly for an adhesion-first reading, a CRISPR screen finds CXCR4 essential INDEPENDENTLY of CXCL12, with deletion eradicating leukaemia without impairing marrow homing - so homing and dependence come apart, and localisation may not be the mechanism doing the work. Oncotarget 2015
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
The CXCR4 antagonist POL5551 has nanomolar activity in pre-B and T ALL cell lines and paediatric ALL primary samples, decreasing CXCR4 antibody binding, blocking SDF-1-alpha-mediated ERK1/2 phosphorylation, inhibiting SDF-1-alpha-induced chemotaxis and reversing stromal-mediated protection from chemotherapy more effectively than plerixafor; it modulates surface expression of adhesion molecules; and it increases sensitivity to cytarabine in a xenograft model of high-risk infant MLL-rearranged ALL.
POL5551, a novel and potent CXCR4 antagonist, enhances sensitivity to chemotherapy in pediatric ALL · Sison EA, Magoon D, Li L, Annesley CE, Romagnoli B, et al. · Oncotarget 2015What this evidence cannot show
Preclinical only; POL5551 has not entered clinical development for ALL and no CXCR4 antagonist has been tested clinically in KMT2A-rearranged infant ALL. The comparison with plerixafor is on the surrogate endpoint of CXCR4 antibody-binding inhibition, not on antileukemic efficacy. Most experiments are in cell lines and general paediatric ALL samples with a single infant MLL-R xenograft model for the in vivo combination result. Published in Oncotarget, whose peer-review standards during this period have been criticised.
Evidence that argues against this (1)
- The clinical translation has not been tested in the genotype this entry is about. No CXCR4-directed trial exists in KMT2A-rearranged or infant leukaemia; the one completed paediatric therapeutic study had a minimum age of 3 years and could not enrol an infant, produced 3 responses in 18 and NONE in ALL. The other paediatric study terminated at n=1 for lack of enrolment. So the axis is untested here rather than refuted, and every termination cites accrual or funding, never safety or efficacy - which means the absence of clinical support carries almost no information either way. British Journal of Haematology 2013
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Among 139 consecutive infants with B-cell precursor ALL on the MLL-Baby protocol, CSF cytospin microscopy found CNS involvement at any level in 50 (36.0%), with a higher incidence in KMT2A-rearranged than KMT2A-germline disease (44.0% vs 15.4%, P = .003); the adverse prognostic impact was confined to the KMT2A-rearranged group (EFS 0.21 for CNS-positive vs 0.48 for CNS-negative, P = .044); CNS-positive infants could not be cured by conventional chemotherapy alone regardless of MRD clearance speed; and the combination of initial CNS negativity plus fusion-gene-transcript MRD negativity identified up to one-third of KMT2A-rearranged infants with disease-free survival above 95%.
Incidence and prognostic value of central nervous system involvement in infants with B-cell precursor acute lymphoblastic leukemia treated according to the MLL-Baby protocol · Popov A, Tsaur G, Permikin Z, Fominikh V, Verzhbitskaya T, et al. · Pediatric Blood & Cancer 2022What this evidence cannot show
CNS status was determined by cytospin microscopy, which is less sensitive than flow cytometry of CSF, so 36% is likely an underestimate and the CNS-negative group is contaminated with occult disease. Single national protocol (MLL-Baby) whose backbone includes all-trans retinoic acid and differs from Interfant, limiting transferability of the outcome figures. The 'above 95% DFS' subgroup is defined post hoc by two favourable variables in a 139-patient cohort and needs prospective validation. 'CNS involvement at any level' pools CNS2 and CNS3 with traumatic taps.
Evidence that argues against this (1)
- CNS involvement in infantile B-ALL is not a fixed property of the age group: across the first year of life the frequency of CNS involvement, hyperleukocytosis, KMT2A rearrangement and lineage switch all fall, with significant breakpoints at 3 and 9 months, so a single pooled percentage for 'infant ALL' obscures a steep age gradient. Knez V et al. 2019
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In the Children's Oncology Group P9407 trial, of 209 analysed cases (199 treatment-eligible) 157 (75.1%) were KMT2A-rearranged; CNS1 (no CSF blasts) occurred in only 71 of 155 (45.8%) KMT2A-rearranged infants versus 32 of 52 (61.5%) KMT2A-germline infants; 5-year EFS by CNS1/CNS2/CNS3 was 54+/-6%, 41+/-8% and 33+/-10% overall (P=0.025), 48+/-7%, 27+/-8% and 30+/-9% in KMT2A-rearranged (P=0.0155) and 53+/-12%, 18+/-10% and 23+/-12% in the KMT2A::AFF1 subgroup (P=0.022); and WBC count at or above 50x10^9/L was present in 123 of 155 (79.4%) KMT2A-rearranged infants, 71 of 77 (92.2%) with AFF1 and 6 of 14 (42.9%) with MLLT3, with WBC differing significantly across partner groups (P<0.0001).
KMT2A partner genes in infant acute lymphoblastic leukemia have prognostic significance and correlate with age, white blood cell count, sex, and central nervous system involvement: a Children's Oncology Group P9407 trial study · Robinson BW, Kairalla JA, Devidas M, Carroll AJ, Harvey RC, et al. · Haematologica 2023What this evidence cannot show
The PubMed record for this article carries no abstract, so title, authors, journal, year, PMID and DOI were verified from PubMed metadata and every number quoted here was read from the publisher's full text at haematologica.org rather than from an abstract. Partner-gene subgroups are small (MLLT3 n=14), so the 42.9% figure has a wide implicit confidence interval. CNS staging uses the CNS1/2/3 system on cytospin morphology. P9407 was a single COG trial with an intensified, shortened regimen, so EFS figures are protocol-specific and not directly comparable with Interfant. The percentage-based partner analyses are secondary and not corrected for multiple comparisons.
Evidence that argues against this (1)
- CNS1 in 45.8% of rearranged versus 61.5% of germline infants is a real difference between groups of 155 and 52, and CNS staging depends on the lumbar puncture: traumatic taps and cytospin technique determine the CNS1/CNS2 boundary, and infants are the hardest patients to tap cleanly. Statistics in Medicine 2006
- E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
In Interfant-06, 651 infants with ALL had 6-year event-free survival of 46.1% (SE 2.1) and overall survival of 58.2% (SE 2.0); risk groups were defined with WBC count as a formal criterion, the high-risk group being KMT2A-rearranged and older than 6 months with WBC count 300x10^9/L or more or a poor prednisone response; the high-risk group had a 6-year EFS of 20.9% (SE 3.4) with the intention to undergo stem cell transplantation, of whom only 46% received it because many had early events; and on prognostic analysis KMT2A rearrangement was the strongest factor for EFS, followed by age, WBC count and prednisone response.
Outcome of Infants Younger Than 1 Year With Acute Lymphoblastic Leukemia Treated With the Interfant-06 Protocol: Results From an International Phase III Randomized Study · Pieters R, De Lorenzo P, Ancliffe P, Aversa LA, Brethon B, et al. · Journal of Clinical Oncology 2019What this evidence cannot show
Interfant-06 did not report CNS involvement frequency in its abstract and the full text is not open access, so this node draws no CNS percentage from it. The WBC 300x10^9/L threshold is a protocol risk-group definition, i.e. an operational cut-off chosen for stratification, and should not be read as a biologically derived inflection point. 6-year EFS differed by 10-12% between West European/North American groups (49.4%) and other countries, so the pooled figures conflate biology with health-system effects. The randomised comparison of myeloid-type versus lymphoid-type consolidation was negative.
Evidence that argues against this (1)
- This is an outcome and risk-stratification result being used in a localisation entry, and the link is indirect: WBC count at diagnosis is a criterion in the risk definition, not a measurement of where disease sits. High WBC is compatible with several distributions of disease, so the trial's risk groups cannot be read as evidence about tissue localisation. It also predates the immunotherapy era that has since changed these numbers. New England Journal of Medicine 2023
- E6established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
In Interfant-99, 482 infants under 1 year treated with a hybrid ALL/AML protocol had a 4-year event-free survival of 47.0% (SE 2.6, 95% CI 41.9-52.1) with 260 (58%) in complete remission at a median follow-up of 38 months, a randomised late intensification course gave no disease-free survival benefit (60.9% vs 57.0% at 4 years, P=0.81), and all types of MLL (KMT2A) rearrangement, very high white blood cell count, age younger than 6 months and poor prednisone-prophase response were independently associated with inferior outcome.
A treatment protocol for infants younger than 1 year with acute lymphoblastic leukaemia (Interfant-99): an observational study and a multicentre randomised trial · Pieters R, Schrappe M, De Lorenzo P, Hann I, De Rossi G, et al. · The Lancet 2007What this evidence cannot show
'Very high white blood cell count' is exactly how the abstract states the prognostic variable; no numeric threshold is given there, so no threshold from Interfant-99 is quoted in this node. 22 countries and 17 study groups over 1999-2005, so supportive care and diagnostics varied. Interfant-99 also reports no CNS involvement frequency in its abstract and its full text is not open access. Outcomes reflect a protocol era now superseded; Interfant-06 did not improve on it.
Evidence that argues against this (1)
- A 4-year EFS of 47% in 482 infants is a trial-cohort figure and belongs to that protocol and era. The randomised late intensification within it is the part with internal validity; the headline survival is a description of what happened to patients enrolled between particular dates under particular supportive care. Leukemia 2021
- E7moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
In a multicentre retrospective cohort of 50 infants diagnosed between 0 and 6 months of life with cutaneous extramedullary myeloid disease (42 AML with cutaneous EMD, 8 isolated cutaneous EMD), born 1990-2018 across Italy, the Netherlands, Switzerland and the United States, KMT2A rearrangement was the most common genetic lesion at 26 of 50 (52%); overall 5-year event-free survival was 66% (95% CI 51-78) and overall survival 75% (95% CI 60-85); CNS involvement was present in 25% of the isolated cutaneous EMD patients; complete spontaneous remission occurred without therapy in two patients; no outcome difference was seen between groups and KMT2A rearrangements were not associated with poorer prognosis; approximately 25-50% of patients with newborn leukemia have cutaneous extramedullary disease and in fewer than 10% skin lesions precede marrow involvement.
Characteristics and treatment of acute myeloid neoplasms with cutaneous involvement in infants up to 6 months of age: A retrospective study · Renaud J, Goemans BF, Locatelli F, Pigazzi M, Redmond S, et al. · Pediatric Blood & Cancer 2024What this evidence cannot show
Fifty patients assembled retrospectively across four countries and 28 birth years, so diagnostics, cytogenetic completeness and therapy are all heterogeneous; the isolated-cutaneous group is only eight patients, making the 25% CNS figure two patients. The 25-50% background rate for cutaneous EMD in newborn leukemia is stated by the authors as context and is not generated by this study. Selection is toward cases that came to specialist attention, which in a condition with documented spontaneous remissions could bias either way. The finding that KMT2A rearrangement is not prognostically adverse here contradicts the marrow literature and is based on 26 KMT2A-r patients.
Evidence that argues against this (1)
- Fifty infants with cutaneous extramedullary MYELOID disease, assembled retrospectively across four countries and 28 years. This is an AML series and the beat's subject is KMT2A-rearranged leukaemia of both lineages, so the localisation pattern it describes cannot be assumed to transfer, and a retrospective multi-decade assembly selects for cases memorable enough to be retrieved. Blood Cells, Molecules and Diseases 2010
- E8moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Across 364 paediatric precursor-B malignancies (210 B-lymphoblastic lymphoma, 154 B-cell precursor ALL), hepatosplenomegaly was present in 80% of BCP-ALL patients versus 2% of BCP-LBL patients, and peripheral blood involvement in 95% versus 2%, making these the two most important discriminators between the leukemic and lymphomatous presentation of the same immature B cell.
Pediatric Precursor B-Cell Lymphoblastic Malignancies: From Extramedullary to Medullary Involvement · Kroeze E, Arias Padilla L, Bakker M, Boer JM, Hagleitner MM, et al. · Cancers (Basel) 2022What this evidence cannot show
This is paediatric BCP-ALL as a whole, not KMT2A-rearranged or infant disease, so the 80% figure is cited for scope and not as a KMT2A-r-specific number; the study's purpose was to contrast ALL with lymphoblastic lymphoma, and hepatosplenomegaly was assessed clinically (palpation/imaging) with no size threshold reported in the abstract. The BCP-ALL arm (n=154) is a comparison group rather than a consecutive leukemia cohort, so it may not be representative of unselected BCP-ALL.
Evidence that argues against this (1)
- The 80%-versus-2% contrast between BCP-ALL and B-lymphoblastic lymphoma is close to definitional: the distinction between leukaemia and lymphoma in precursor-B disease is drawn on marrow blast percentage and peripheral involvement, so a cohort split on that criterion will differ on hepatosplenomegaly and blood involvement by construction. Statistics in Medicine 2006
- E9moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
In a fully murine immunocompetent KMT2A-AFF1+ mixed-lineage infant leukemia model, bone-marrow-derived leukemia propagating cells engrafted the CNS unequally with lineage-negative cKit+ common lymphoid progenitors consistently underrepresented; CNS-derived leukemia propagating cells showed reduced systemic repopulation capacity with many recipients showing stable long-term engraftment without overt leukemia; and transcriptomic profiling of matched CNS- and bone-marrow-derived cells showed CNS enrichment of hypoxia, lipid and cholesterol homeostasis and inflammatory signalling pathways, upregulation of lipid and fatty-acid metabolic programs in the CNS-adapted subsets, and increased expression of T-cell immune-modulation genes.
Mixed-lineage leukemia cells undergo unique adaptations in the CNS niche · Duguid A, Malouf C, Nitsche L, Halsey C, Ottersbach K · Experimental Hematology 2026What this evidence cannot show
A single murine model of KMT2A-AFF1 mixed-lineage leukemia; the murine CNS is small and its meningeal vasculature and CSF dynamics differ substantially from an infant's, and the model's mixed-lineage phenotype may not represent pro-B KMT2A::AFF1 infant ALL. Transcriptomic pathway enrichment is descriptive and no metabolic intervention was tested. The counter-intuitive finding that CNS-derived cells are less systemically aggressive is exactly the sort of result most in need of independent replication and has none yet. Recent publication.
Evidence that argues against this (1)
- A fully murine immunocompetent model is the right design for a CNS question and the result is a compartment comparison within mouse haematopoiesis. Unequal CNS engraftment by sorted populations shows which murine progenitors reach the CNS in that model; human CNS disease at diagnosis arises before any transplantation. Oncofusion knowledge graph (internal analysis) 2026
- E10strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In the Project EVOLVE post-immunotherapy lineage-switch cohort of 70 patients, CNS disease was present in 10 (14.3%) and non-CNS extramedullary disease in 5 (7.1%) at the time of the most proximal immunotherapy.
Project EVOLVE: an international analysis of postimmunotherapy lineage switch, an emergent form of relapse in leukemia · Silbert SK, Rankin AW, Hoang CN, Semchenkova A, Myers RM, et al. · Blood 2025What this evidence cannot show
These percentages describe a heavily pre-treated, relapsed, solicited cohort of patients who went on to undergo lineage switch, spanning 12 days to 76.5 years at initial diagnosis; they are not diagnosis-time extramedullary rates for KMT2A-rearranged leukemia and must not be read as such. No denominator for the parent population exists.
Evidence that argues against this (1)
- CNS disease in 14.3% at the time of the most proximal immunotherapy is measured in a cohort assembled BECAUSE it switched lineage, so it describes CNS involvement among switchers rather than among KMT2A-rearranged patients generally. Journal of Clinical Oncology 2025
- E11weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper
A patient with refractory KMT2A/AFF1-rearranged B-ALL developed, during the fourth cycle of blinatumomab, an extradural thoracic mass causing spinal cord compression that was biopsy-proven mixed-phenotype myeloid sarcoma carrying the same KMT2A rearrangement, with concurrent mixed-phenotype acute leukemia in the marrow.
Blinatumomab-Related Lineage Switch of KMT2A/AFF1-Rearranged B-Lymphoblastic Leukemia to B/Myeloid Mixed-Phenotype Acute Leukemia and Myeloid Sarcoma Causing Spinal Cord Compression · Fan X, Gupta A, Dabrowski D, Rusella M, Zhai QJ, et al. · Case Reports in Hematology 2025What this evidence cannot show
A single case report in a 19-year-old, not an infant, and the patient was lost to follow-up. Cited purely to document that extramedullary relapse at an unusual site can be the presenting manifestation of lineage switch; it supports no frequency claim.
Evidence that argues against this (1)
- One patient. A biopsy-proven mixed-phenotype myeloid sarcoma carrying the same rearrangement during blinatumomab is a well-documented event and cannot establish a rate, a mechanism or a preferential site. Blood Cells, Molecules and Diseases 2010
- E12moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
A multi-institutional review of 54 cases of infantile B-lymphoblastic leukemia across four medical institutions, divided into congenital and non-congenital groups with multiple age subgroups, found decreasing trends across the first year of life in hyperleukocytosis, central nervous system involvement, KMT2A rearrangement, lineage switch and mortality, against increasing trends in CD10 expression and non-KMT2A abnormalities, with statistically significant differences emerging at 3 and at 9 months, such that the last trimester of infantile B-ALL essentially merges with paediatric B-ALL.
Clinicopathologic and genetic spectrum of infantile B-lymphoblastic leukemia: a multi-institutional study · Knez V, Liu X, Schowinsky J, Pan Z, Wang D, et al. · Leukemia & Lymphoma 2019What this evidence cannot show
Fifty-four cases split across congenital/non-congenital and multiple age subgroups leaves very small cells for the 3-month and 9-month comparisons, so the breakpoints should be treated as hypothesis-generating. The trends are reported qualitatively in the abstract without percentages, so no numeric value from this study is quoted. Retrospective pathology review across four institutions with differing diagnostic panels and eras.
Evidence that argues against this (1)
- Fifty-four cases across four institutions, divided into congenital and non-congenital groups and then into multiple age subgroups, which leaves very few patients per cell. "Decreasing trends across the first year of life" derived from such a subdivision is the shape of analysis most likely to find a trend by chance. Statistics in Medicine 2006
the CNS as a state rather than a location; this entry supplies the verified author list for PMID 41371538, which the marrow-niche ledger recorded as unverified
hyperleukocytosis, organomegaly, CNS and extramedullary involvement as the presenting anatomical picture
NG2 is further upregulated on blasts infiltrating extramedullary sites and CNS, so the invasive phenotype is a property of this state
the marrow niche the leukaemia occupies and displaces
- related to →Where KMT2A and its fusions sit inside the nucleus — SNL1/SNL2, nuclear speckles and the condensate question
companion node: localization at the subnuclear rather than the anatomical scale
defines the marrow compartment that entry records KMT2A-r blasts occupying
identifies the stromal source of the CXCL12 that entry records as the best-characterised niche dependency
supplies the mechanistic detail, and the CXCL12-independent counter-result, behind the niche dependency recorded there