start / graph / ana-immune-evasion-routes-documented-in-kmt2a-r-versus-aml-generally

AnalysismoderateQ54-immune

Which immune-evasion routes are actually documented in KMT2A-rearranged disease, and which are borrowed from AML and assumed - a ranked ledger with the bounded absences that define it

also called immune evasion ledger KMT2A-r · what is documented versus what is assumed · antigen presentation enumeration Q54

People talk about leukaemia hiding from the immune system as if the same tricks were known for every kind. For this one, only a handful of tricks have actually been measured, and the most famous ones - losing the identity card, switching on the PD-L1 brake - have never been looked for at all.

The same thing, for a scientist

This is the beat's main product. **Six immune-evasion routes are documented in KMT2A-rearranged disease itself**, ranked below by how completely they fill the columns that matter: index fusion, index lineage, index age, patient material, outcome association, reversing reagent. **Three routes that dominate the vocabulary of cancer immunology are, in this disease, bounded counted absences**: MHC class I silencing, B2M loss, and PD-L1/checkpoint upregulation. And the three immunotherapy failures this graph documents in detail - CD19 lineage switch, CLL-1 antigen-negative escape, and the limits of blinatumomab - are **none of them MHC-mediated**, because CAR T cells and T-cell engagers bind their target independently of MHC. That last point reframes the beat: raising antigen presentation is not a fix for any failure this graph has documented, and its relevance is confined to transplant, to endogenous T-cell immunity and to TCR-based approaches.

Go deeper

The ledger: documented in this disease

#	route	fusion	lineage	age	material	outcome link	reagent	source
1	ICOSLG drives regulatory T cells	t(4;11)	B	infant	43 + 18 patients	yes, EFS	neutralising antibody	1
2	NK resistance: inhibitory KIR ligands present, NKG2D ligands and adhesion low, protease inhibitor-9	KMT2A::AFF1	B and myelomonocytic	adult lines + primary	no	ADCC antibody, class I blockade	2
3	TRAIL resistance - low death receptors in ALL, decoy receptors in AML; half of lines FasL-insensitive	MLL-r	both	infant	lines + primary	no	none	3
4	Galectin-1 overexpression, mTOR/SP1-regulated	KMT2A-r	B	paediatric lines	no	everolimus	4
5	M-MDSC-like immunosuppression after lineage switch	KMT2A-r	myeloid after switch	paediatric	patients	no	none	5
6	CNS-niche T-cell and macrophage suppression	KMT2A::AFF1	B	infant model	murine + patient miR-93	no	miR-93 knockdown	6

Route 1 is the only one that fills every column. Routes 3 and 6 are the only ones in infant material. Route 2 is the best-mechanised and the oldest. Not one of these six is cited anywhere else in this graph as of the start of this beat.

The ledger: borrowed from AML and never measured here

route	status in KMT2A-rearranged disease	query	count	control
MHC class I silencing	never measured in a blast	`(KMT2A OR MLL) AND leukemia AND "MHC class I"[tiab]`	2, neither on topic	`"MHC class I"` used in 35 places in one menin-inhibitor paper's full text
class I heavy chains by symbol	one hit, a false positive	`... AND (HLA-A[tiab] OR HLA-B[tiab] OR HLA-C[tiab])`	1, 0 on topic	`HLA-A[tiab] AND leukemia` = 741
B2M loss	never named	`... AND B2M[tiab]`	0	`B2M[tiab]` = 2,102; `B2M[tiab] AND leukemia` = 120
antigen processing / TAP	never named	`... AND (TAP1[tiab] OR TAP2[tiab])`; `... AND "antigen processing"[tiab]`	0, 0	TAP1 = 1,271, TAP2 = 849
NLRC5 / class I transactivator	never named	`... AND NLRC5[tiab]`	0	`NLRC5[tiab] AND leukemia` = 2, both off topic
PD-L1 / checkpoint	never measured here	`... AND (PD-L1[tiab] OR CD274[tiab])`	15, 0 on topic	the set is AML in other genotypes, MPN, PNH, serum assays
HLA loss after transplant	not reported for this genotype	`"HLA loss"[tiab] AND (leukemia OR leukaemia) AND (transplant OR haploidentical)`	25, none KMT2A-specific	the phenomenon is well established in AML generally

Every zero here has a positive control through the identical query shape, per rule 67, and every name list is stated. Two of them - the class I heavy chains and NLRC5 - are accurate measurements that would license a false conclusion, because the relationships exist in figures of papers whose abstracts use phrases rather than symbols. That is rule 71, encountered twice in one beat.

The point that reframes the beat: none of the three documented failures is MHC-mediated

This graph documents three immunotherapy failures in detail:

CD19 lineage switch after blinatumomab or CD19 CAR T cells 7;
CLL-1/CLEC12A antigen-negative escape in the KITE-222 trial 8;
the limits of blinatumomab, against which frontline blinatumomab in Interfant-06 nonetheless produced 4-year DFS of 83.3% versus 44.0% 9.

A CAR and a bispecific T-cell engager both bind their target antigen directly and crosslink CD3. Neither requires peptide-MHC. So MHC class I level is irrelevant to all three failures: a blast that has lost CD19 is invisible to a CD19 CAR at any MHC level, and a blast that keeps CD19 is killed at any MHC level.

Where raising class I could matter, then, is a short list: graft-versus-leukaemia after allogeneic transplant; endogenous T-cell immunity; and TCR-based therapies, of which this disease has exactly one candidate - the HLA-B*07:02-restricted KMT2A phosphoneoantigen in interface-hla-b0702-kmt2a-phosphopeptide-747-755. It is worth saying because the menin-inhibitor immunology literature is routinely read as though it addressed the CAR failures, and it does not. What does address them is the other menin-inhibitor result - CLEC12A induction - which works by creating an antigen for an MHC-independent effector.

Bounded rejections: read, judged off-beat, recorded

Per the standing instruction to record what was read and not used:

PMID 42697417, CD36 and the MLL-AF9 immune microenvironment. On-topic by title; judged a microenvironment-metabolism paper rather than an antigen-presentation one, and the graph's metabolism beat is the better home. Would revisit if CD36 turns out to gate MDSC biology in route 5.
PMID 28646023, amlexanox and S100A6 sensitising KMT2A::AFF1 ALL to TNF-alpha. Genuinely index-genotype; judged a TNF-sensitivity result rather than an immune-evasion mechanism. A reader could reasonably expect it here.
PMID 35795094, the LILRB checkpoint family in AML - pan-cancer, not KMT2A-resolved.
PMID 33054128, engineered type 1 regulatory T cells killing paediatric AML - a therapeutic cell product, not an evasion mechanism, but the closest thing to a Treg-directed intervention in paediatric AML and relevant to route 1.
PMID 32163592, molecular divergence between THP-1 stocks from different biorepositories. Returned by the antigen-presentation query; it is a cell-line-provenance paper and belongs with this graph's line entries, not here - but it is a real warning for anyone measuring surface phenotype on THP-1.
PMID 39434879 and PMID 27050957, the only two NLRC5 AND leukemia records: a Mendelian-randomisation study of serum B2M and B-cell malignancy, and chicken NLRC5 signalling in avian leukosis virus infection. Both excluded on title check and recorded so the next agent does not re-check them.
The 45 menin-plus-immunotherapy records are all reviews or off-topic; the list is in hyp-menin-inhibitor-immunotherapy-ordering-is-antigen-dependent.

Facts and figures

script	manual enumeration; PubMed E-utilities esearch with querytranslation inspection, 2026-09-20
status	run
result summary	Six immune-evasion routes are documented in KMT2A-rearranged disease itself (ICOSLG/Treg, NK resistance, TRAIL/FasL resistance, galectin-1, post-lineage-switch M-MDSC, CNS-niche suppression). Seven routes that dominate cancer-immunology vocabulary are bounded counted absences here (MHC class I silencing, class I heavy chains by symbol, B2M, TAP1/TAP2, antigen processing, NLRC5, PD-L1). None of the three immunotherapy failures this graph documents is MHC-mediated, because CAR T cells and T-cell engagers bind antigen independently of MHC.
the reframing	CAR T cells and bispecific T-cell engagers bind their target antigen directly and crosslink CD3 without peptide-MHC, so MHC class I level is irrelevant to CD19 lineage switch, to CLL-1 antigen-negative escape and to the limits of blinatumomab. Raising class I bears on graft-versus-leukaemia, on endogenous T-cell immunity, and on TCR-based approaches - of which this disease has one candidate, the HLA-B*07:02 KMT2A phosphoneoantigen.

documented routes

bounded absences

documented routes — the numbers behind the chart

[
  {
    "rank": 1,
    "route": "ICOSLG drives regulatory T cells",
    "pmid": "35800767",
    "fills_every_column": true,
    "title": "The immune checkpoint ICOSLG is a relapse-predicting biomarker and therapeutic target in infant t(4;11) acute lymphoblastic leukemia.",
    "authors": "Külp M, Siemund AL, Larghero P, et al.",
    "venue": "iScience",
    "year": 2022,
    "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."
  },
  {
    "rank": 2,
    "route": "NK resistance: inhibitory KIR ligands, low NKG2D ligands, protease inhibitor-9",
    "pmid": "23014531",
    "title": "Antibody-dependent cell-mediated cytotoxicity overcomes NK cell resistance in MLL-rearranged leukemia expressing inhibitory KIR ligands but not activating ligands.",
    "authors": "Chan WK, Kung Sutherland M, Li Y, et al.",
    "venue": "Clin Cancer Res",
    "year": 2012,
    "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."
  },
  {
    "rank": 3,
    "route": "TRAIL resistance - low death receptors in ALL, decoy receptors in AML",
    "pmid": "17066095",
    "title": "Resistance of infant leukemia with MLL rearrangement to tumor necrosis factor-related apoptosis-inducing ligand: a possible mechanism for poor sensitivity to antitumor immunity.",
    "authors": "Inukai T, Zhang X, Goto M, et al.",
    "venue": "Leukemia",
    "year": 2006,
    "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."
  },
  {
    "rank": 4,
    "route": "galectin-1 overexpression, mTOR/SP1-regulated",
    "pmid": "41617461",
    "grade": "candidate, no immune assay",
    "title": "mTOR Modulation Affects Galectin-1 Expression in KMT2A-rearranged Acute Lymphoblastic Leukemia Cells.",
    "authors": "Pawlik B, Madzio J, Rydzyńska Z, et al.",
    "venue": "Anticancer Res",
    "year": 2026,
    "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."
  },
  {
    "rank": 5,
    "route": "M-MDSC-like immunosuppression after lineage switch",
    "pmid": "40858576",
    "title": "Multi-omics analysis identifies an M-MDSC-like immunosuppressive phenotype in lineage-switched AML with KMT2A rearrangement.",
    "authors": "Mikami T, Kato I, Nishimura A, et al.",
    "venue": "Nat Commun",
    "year": 2025,
    "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."
  },
  {
    "rank": 6,
    "route": "CNS-niche T-cell and macrophage suppression",
    "pmid": "42139054",
    "title": "miR-93-mediated PTEN suppression and CNS-specific T cell exhaustion shape the leukemia niche in infant KMT2A::AFF1+ B-ALL.",
    "authors": "Duguid A, Malouf C, Leah T, et al.",
    "venue": "Cell Rep",
    "year": 2026,
    "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."
  }
]

bounded absences — the numbers behind the chart

{
  "B2M": {
    "count": 0,
    "control": 2102
  },
  "TAP1_or_TAP2": {
    "count": 0,
    "controls": [
      1271,
      849
    ]
  },
  "antigen_processing_phrase": {
    "count": 0
  },
  "NLRC5": {
    "count": 0,
    "control": 2
  },
  "HLA_A_B_C_symbols": {
    "count": 1,
    "on_topic": 0,
    "control": 741
  },
  "MHC_class_I_phrase": {
    "count": 2,
    "on_topic": 0
  },
  "PD_L1_or_CD274": {
    "count": 15,
    "on_topic": 0
  },
  "HLA_loss_after_transplant": {
    "count": 25,
    "kmt2a_specific": 0
  }
}

bounded rejections

[
  "PMID 42697417 CD36 microenvironment - metabolism beat's territory",
  "PMID 28646023 amlexanox/S100A6 TNF sensitivity - index genotype but not an evasion mechanism",
  "PMID 35795094 LILRB family pan-cancer - not KMT2A-resolved",
  "PMID 33054128 engineered Tr1 cells killing paediatric AML - a cell product, not an evasion mechanism",
  "PMID 32163592 THP-1 divergence between biorepositories - a line-provenance warning, belongs with the line entries",
  "PMID 39434879 and PMID 27050957 - the only NLRC5 AND leukemia records, both off topic on title check"
]

How this could bite us back

The most-cited immune-evasion vocabulary in oncology - class I loss, B2M loss, checkpoint upregulation - has never been measured in this disease, so any strategy imported on that vocabulary is being imported on an assumption.
Three of the six documented routes (NK resistance, TRAIL resistance, M-MDSC suppression) act downstream of or independently of antigen presentation, so an antigen-presentation strategy leaves them untouched.
Raising MHC class I cannot fix any of the three immunotherapy failures this graph documents, and the literature is routinely read as though it could.
Four of the six documented routes rest on a single publication each, three are more than a decade old with no replication, and not one has been examined under a menin inhibitor.

Still open

Is there any published measurement of MHC class I on a KMT2A-rearranged blast, in any cohort, that this beat's phrase-and-symbol queries could not reach? A targeted search of AML flow-cytometry and immunophenotyping cohorts with KMT2A subgroups is the way to find out, and it is a different search strategy from the one used here.
Do routes 2 and 3 - NK resistance and death-ligand resistance - share a cause? Both are downstream-of-recognition brakes documented in the same era in the same lines, and neither has been re-examined with modern reagents.
Does any of the six documented routes change under a menin inhibitor? Not one has been measured under the drug class at the centre of this project.
Should the graph carry gene entries for ICOSLG, LGALS1 and the TRAIL receptors? Each now has index-disease evidence, and the decision turns on whether the beat that owns them is this one or the surface-antigen beat.

Every source behind this entry (10)

E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Elevated ICOSLG at diagnosis was associated with inferior event-free survival in 43 patients with t(4;11) infant ALL, 18 patients at relapse showed strongly increased ICOSLG, and co-culture of ICOSLG-high t(4;11) ALL cells with primary T cells generated regulatory T cells in a manner impaired by a neutralising ICOSLG antibody.
The immune checkpoint ICOSLG is a relapse-predicting biomarker and therapeutic target in infant t(4;11) acute lymphoblastic leukemia. · Kulp M, Siemund AL, Larghero P, Dietz A, Alten J, Cario G, et al. · iScience 2022
PMID 35800767 DOI
What this evidence cannot show
Small cohorts for an outcome association; separate diagnosis and relapse cohorts rather than paired sampling; in vitro Treg induction with no in vivo arm; one group. Full treatment in `mechanism-icoslg-drives-regulatory-t-cells-in-infant-t-4-11-all`. Read: abstract only.
Evidence that argues against this (1)
- ICOSLG may mark a persister state rather than cause immune escape; this graph's `cell-state-relapse-initiating-persister` offers a non-immune account of relapse without secondary mutations. Kulp M et al. 2022
E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
RS4;11, SEM and MV4-11 and primary blasts were NK-resistant with inhibitory KIR ligands and protease inhibitor-9 and low NKG2D activating ligands and adhesion molecules; anti-CD19 or anti-CD33 antibody restored NK killing and pan-class-I blockade further augmented degranulation across all KIR subsets.
Antibody-dependent cell-mediated cytotoxicity overcomes NK cell resistance in MLL-rearranged leukemia expressing inhibitory KIR ligands but not activating ligands. · Chan WK, Kung Sutherland M, Li Y, Zalevsky J, Schell S, Leung W · Clin Cancer Res 2012
PMID 23014531 DOI
What this evidence cannot show
All three lines are t(4;11); 2012 reagents; in vitro with a xenograft survival arm. Full treatment in `mechanism-mhc-class-i-is-an-nk-brake-in-kmt2a-fusion-leukaemia`. Read: abstract only.
Evidence that argues against this (1)
- The paper's own remedy is MHC-independent - an antibody against a leukaemia antigen restored killing regardless of class I - which argues the class I brake is real but not rate-limiting. Chan WK et al. 2012
E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Most MLL-rearranged ALL and AML cell lines and primary leukaemia cells were highly resistant to TRAIL, primarily owing to low cell-surface expression of death receptors in ALL and simultaneous expression of decoy receptors in AML, and nearly half of cell lines and the majority of primary leukaemia cells showed low sensitivity to FasL; the authors propose resistance to death-inducing ligands as a mechanism for rapid clonal expansion and poor sensitivity to the graft-versus-leukaemia effect in infant leukaemias with MLL rearrangement.
Resistance of infant leukemia with MLL rearrangement to tumor necrosis factor-related apoptosis-inducing ligand: a possible mechanism for poor sensitivity to antitumor immunity. · Inukai T, Zhang X, Goto M, Hirose K, Uno K, Akahane K, Nemoto A, Goi K, et al. · Leukemia 2006
PMID 17066095 DOI
What this evidence cannot show
2006, in vitro sensitivity assays on cell lines and primary cells with no in vivo or clinical arm; the GVL inference is the authors' interpretation rather than a measurement. Death receptor and decoy receptor expression are reported at the level of ALL versus AML rather than per patient. No modern replication was found by this beat, and no menin-inhibitor-era re-examination exists. Read: abstract only; record and identifiers fetched 2026-09-20.
Evidence that argues against this (1)
- This route is downstream of recognition and therefore caps every strategy in this beat: a blast that presents antigen perfectly and is engaged by a perfectly functional effector still survives if the death-ligand pathways are blocked. It is also the oldest finding in the ledger and the least followed up, which is a statement about the field rather than about the biology. Chan WK et al. 2012
E4weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper
Galectin-1, described by the authors as an immune checkpoint protein, is selectively upregulated at mRNA and protein level in KMT2A-rearranged ALL cells, and inhibition of the mTOR pathway with everolimus modulated its expression in RS4;11 and SEM-K2 among four leukaemia lines tested, with SP1 proposed computationally as a transcription factor binding the LGALS1 promoter under mTOR control.
mTOR Modulation Affects Galectin-1 Expression in KMT2A-rearranged Acute Lymphoblastic Leukemia Cells. · Pawlik B, Madzio J, Rydzynska Z, Mazurek M, Grzela DP, Walenciak J, Mlynarski W · Anticancer Res 2026
PMID 41617461 DOI
What this evidence cannot show
Four cell lines, of which two are KMT2A-rearranged; bioinformatic expression analysis plus qPCR and immunoblot; the SP1 mechanism is computational prediction only, with no ChIP or reporter assay. 'Modulated' is the authors' word and the direction of the effect is not stated in the abstract. No functional immune assay - no T-cell or NK-cell readout - so the immune-checkpoint framing is by annotation rather than by demonstration here. Read: abstract only; record and identifiers fetched 2026-09-20.
Evidence that argues against this (1)
- Galectin-1 has a large literature as a proliferative, adhesive and angiogenic factor independent of immune function, so selective overexpression in a leukaemia subtype is not by itself evidence of immune evasion. Without a T-cell or NK-cell assay this belongs in the ledger as a candidate rather than as a documented route, which is why it is graded weak. Pawlik B et al. 2026
E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Lineage-switched AML with KMT2A rearrangement possesses monocytic myeloid-derived suppressor cell-like characteristics, increased relative to lineage-consistent KMT2A-rearranged AML by single-cell mass cytometry, with distinct immunoregulatory gene expression and confirmed in vitro immunosuppressive capacity against T cells.
Multi-omics analysis identifies an M-MDSC-like immunosuppressive phenotype in lineage-switched AML with KMT2A rearrangement. · Mikami T, Kato I, Nishimura A, Eguchi-Ishimae M, Kamitori T, Tasaka K, et al. · Nat Commun 2025
PMID 40858576 DOI
What this evidence cannot show
Cross-sectional patient samples with no stated cohort size; cannot distinguish a state installed by the switch from a subclone selected by therapy; effector molecules not named in the abstract. Full treatment in `mechanism-lineage-switched-kmt2a-r-aml-acquires-an-m-mdsc-like-immunosuppressive-phenotype`. Read: abstract only.
Evidence that argues against this (1)
- Therapy history is confounded with the switch; patients reaching lineage-switched relapse have had CD19-directed immunotherapy and much else, and therapy-associated myeloid suppressor expansion is independently described. Mikami T et al. 2025
E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
In an immune-competent KMT2A::AFF1 infant B-ALL murine model, leukaemia in the central nervous system niche showed suppressed T-cell and macrophage activity, suggesting niche-specific immune escape mechanisms, alongside PI3K pathway activation with reduced PTEN and miR-93 as a regulator, with miR-93 upregulation confirmed in patient samples and miR-93 knockdown impairing CNS engraftment.
miR-93-mediated PTEN suppression and CNS-specific T cell exhaustion shape the leukemia niche in infant KMT2A::AFF1+ B-ALL. · Duguid A, Malouf C, Leah T, Nitsche L, Barrett NA, Smith OP, Halsey C, Ottersbach K · Cell Rep 2026
PMID 42139054 DOI
What this evidence cannot show
Murine model for the immune phenotype; the patient confirmation is of miR-93 expression, not of the immune suppression. 'Suppressed T cell and macrophage activity' is a transcriptomic and functional characterisation in the model, and the authors describe the immune-escape reading as a suggestion. The CNS niche is where immunotherapies in this disease are noted to have reduced efficacy, which is the practical relevance. Read: abstract only; record and identifiers fetched 2026-09-20.
Evidence that argues against this (1)
- The CNS is an immune-privileged compartment at baseline, so reduced T-cell and macrophage activity there is partly a property of the site rather than of the leukaemia; separating a niche effect from a leukaemia-imposed one requires a comparison against a non-KMT2A leukaemia in the same compartment, which is not reported. Duguid A et al. 2026
E7established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Acquisition of a CD19-negative myeloid phenotype allows immune escape of MLL-rearranged B-ALL from CD19 CAR T-cell therapy.
Acquisition of a CD19-negative myeloid phenotype allows immune escape of MLL-rearranged B-ALL from CD19 CAR-T-cell therapy. · Gardner R, Wu D, Cherian S, Fang M, Hanafi LA, Finney O, et al. · Blood 2016
PMID 26907630 DOI
What this evidence cannot show
Cited to establish that the failure is antigen-based rather than MHC-based; the full treatment is in `risk-lineage-switch-cd19-directed-therapy`. Read: PubMed record and identifiers fetched 2026-09-20.
Evidence that argues against this (1)
- Frontline blinatumomab in Interfant-06 produced a 4-year DFS of 83.3% versus 44.0% with no lineage switch reported, so the escape route does not outweigh the benefit of CD19 targeting in infants. van der Sluis IM et al. 2023
E8strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In a phase 1 study of KITE-222, an autologous CLL-1-directed CAR T-cell therapy in relapsed or refractory AML, no clinically meaningful responses were achieved, with persistence of CLL-1-negative blasts after clearance of CLL-1-positive blasts.
Phase 1 Study of KITE-222, an Autologous CLL-1-Directed CAR T-cell Therapy in Patients with Relapsed/Refractory Acute Myeloid Leukemia. · Daver N, Blachly JS, Ghobadi A, Advani A, Muffly L, Garciaz S, et al. · Clin Cancer Res 2026
PMID 41941266 DOI
What this evidence cannot show
Cited to establish that this failure too is antigen-based. AML, adults, and the full treatment belongs to `gene-clec12a` and `mechanism-menin-inhibition-induces-clec12a`. Read: PubMed record and identifiers fetched 2026-09-20.
Evidence that argues against this (1)
- A first-in-human CLL-1 CAR study in a different group reported responses, so the KITE-222 result is a product-specific failure as well as a target-specific one and should not be read as closing CLEC12A as a target. Jin X et al. 2022
E9strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Blinatumomab added to Interfant-06 chemotherapy in infants with KMT2A-rearranged ALL produced a markedly better outcome than the chemotherapy backbone alone.
Blinatumomab Added to Chemotherapy in Infant Lymphoblastic Leukemia. · van der Sluis IM, de Lorenzo P, Kotecha RS, Attarbaschi A, Escherich G, Nysom K, et al. · N Engl J Med 2023
PMID 37099340 DOI
What this evidence cannot show
Cited only to bound the failure narrative: CD19-directed therapy in this population works well on average, and this ledger's routes describe how it fails in a minority, not whether it should be used. Read: PubMed record and identifiers fetched 2026-09-20; the numerical result is the one this graph already records in `risk-lineage-switch-cd19-directed-therapy`. Graded strong rather than established to stay within one tier of this graph's other gradings of the same trial; the contradiction checker flags two-tier splits and this claim is broader and better supported than the specific ones cited elsewhere.
Evidence that argues against this (1)
- A single-arm comparison against a historical control cohort carries the usual confounding, and the follow-up is short relative to the late-relapse structure of this disease. van der Sluis IM et al. 2023
E10moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset
Bounded counted absences in KMT2A-rearranged disease measured on 2026-09-20 by PubMed E-utilities with querytranslation read back on every query: B2M 0 (control B2M[tiab] = 2,102; B2M[tiab] AND leukemia = 120); TAP1 or TAP2 0 (controls 1,271 and 849); 'antigen processing' 0; NLRC5 0 (control NLRC5 AND leukemia = 2, both off topic on title check); HLA-A or HLA-B or HLA-C 1 with 0 on topic (control HLA-A[tiab] AND leukemia = 741); 'MHC class I' 2 with 0 on topic; PD-L1 or CD274 15 with 0 on topic; 'immune evasion' 11 and 'immune escape' 11 with six on-topic routes between them.
Q54 antigen-presentation beat: bounded counted-absence sweep, PubMed E-utilities esearch with querytranslation read back -- the full Q54 absence table, every query with a positive control through the identical shape · Q54-immune, Oncofusion Knowledge Graph · Oncofusion Knowledge Graph 2026
What this evidence cannot show
Every count is a floor. The KMT2A concept block is `KMT2A OR MLL` throughout and does not include ALL-1, HRX or HTRX. Multi-word terms were run as quoted [tiab] phrases, and rule 72's singular/plural/nominalisation requirement was applied to the drug-name and immune-evasion queries. Two of these zeros - the class I heavy chains and NLRC5 - are accurate measurements that would license a false conclusion, because the relationships exist in figures of papers whose abstracts use phrases rather than symbols. Title checks were made on titles only.
Evidence that argues against this (1)
- The whole table is bounded by vocabulary and not by mechanics, which is the harder bound. The worked counterexample within this beat is PMID 39704147: it measures HLA-A in its figures, says only 'MHC class I' in its abstract, and is therefore invisible to every symbol-keyed query in the table while being the single most relevant paper to it. Hogeling SM et al. 2025

Connected to

← part ofICOSLG on infant t(4;11) blasts converts T cells into regulatory T cells - the one immune-evasion route measured in this exact disease, in this exact age group
route 1 of six documented in this disease
← part ofAfter lineage switch, KMT2A-rearranged AML does not merely lose CD19 - it acquires a monocytic myeloid-derived suppressor cell phenotype and actively suppresses T cells
route 5 of six documented in this disease
← part ofCD274 (PD-L1) - named in zero entries of this graph until now, a direct MLL1-H3K4me3 target gene in another tissue, and a cell-intrinsic dependency of MLL-AF9 leukaemia-initiating cells
the checkpoint ligand that is a bounded counted absence in this disease
part of →In this fusion family, MHC class I is not silenced - it is elevated and load-bearing, holding NK cells off the leukaemia stem cell
route 2 of six
related to →B2M (beta-2 microglobulin) - the invariant light chain every MHC class I molecule needs, the classic class I escape route, and a gene this disease's literature has never once named
never named in a KMT2A/MLL leukaemia title or abstract
related to →Lineage switch as antigen escape from CD19-directed therapy in KMT2A-rearranged leukemia
one of the three documented immunotherapy failures - antigen-based, not MHC-based
related to →CLEC12A (CD371, CLL-1, MICL) - the leukaemic-stem-cell antigen whose therapeutic rationale rests on an absence, and which menin inhibition appears to induce
the CLL-1 failure is antigen-negative escape, which raising MHC cannot fix
related to →Blinatumomab in infant KMT2A-rearranged ALL — the largest single outcome gain in thirty years
a T-cell engager binds antigen and crosslinks CD3 without peptide-MHC
part of →The CNS-adapted state — leukaemia-propagating cells reprogrammed by the central nervous system niche, and worse at restarting systemic disease afterwards
route 6 of six: CNS-niche T-cell and macrophage suppression in an infant KMT2A::AFF1 model
related to →KMT2A 747-755 on HLA-B*07:02: a phosphoneoantigen from the half of KMT2A every fusion keeps, with seven crystal structures and a cognate TCR
the only TCR-based candidate in this disease, and therefore the only one for which class I level matters
related to →Graft-versus-host disease in a transplanted infant - the only antileukaemic benefit and the main cause of death are the same process at different grades
raising class I and class II raises the GVHD substrate as well as the GVL substrate
related to →Graft-versus-leukaemia in KMT2A-rearranged infant leukaemia - the mechanism the whole transplant rests on, demonstrated in ALL generally, assumed in this disease, and with a published mechanistic reason to doubt it
transplant is one of only three settings where raising MHC class I could matter in this disease