After lineage switch, KMT2A-rearranged AML does not merely lose CD19 - it acquires a monocytic myeloid-derived suppressor cell phenotype and actively suppresses T cells
also called M-MDSC-like lineage-switched AML · immunosuppression after lineage switch · lineage switch is not only antigen loss
When this leukaemia escapes CD19-targeted therapy by turning into a myeloid leukaemia, it does not just drop the target. The cells it becomes are the kind that shut T cells down, which is a second reason the immune therapy stops working.
The same thing, for a scientist
This graph documents lineage switch thoroughly as **antigen escape** - `risk-lineage-switch-cd19-directed-therapy` and `mechanism-lineage-switch-kmt2a` both frame it that way, correctly. Mikami et al. add a second axis. Multi-omics of patient samples shows that **lineage-switched AML cells with KMT2A rearrangement possess monocytic myeloid-derived suppressor cell (M-MDSC)-like characteristics**; single-cell mass cytometry shows the M-MDSC-like population **increased in lineage-switched AML relative to lineage-consistent KMT2A-rearranged AML**; single-cell transcriptomics identifies distinct immunoregulatory gene expression in that population; and in vitro assays confirm **immunosuppressive capacity against T cells, analogous to MDSCs** [E1]. **So the escape is two-layered: the target disappears and the effector is suppressed.** The comparator is the right one - lineage-consistent KMT2A-rearranged AML rather than healthy marrow - which is what makes this a statement about the switch rather than about myeloid identity in general.
Why the comparator is the whole result
MDSC-like features are common in myeloid malignancy, so the interesting question is never 'does this AML look myeloid-suppressive' but 'does it look more so than a matched AML that did not switch'. Mikami's design answers that: lineage-switched KMT2A-rearranged AML versus lineage-consistent KMT2A-rearranged AML 1. Same fusion, same lineage at the time of measurement, different history.
That design also bounds the claim. It does not establish whether the M-MDSC-like state is:
- installed by the switch - a consequence of the transdifferentiation programme;
- selected by the therapy - CD19-directed pressure selecting a pre-existing suppressive subclone, which is the model
hyp-lineage-switch-is-competitive-release-of-preexisting-cloneproposes for the switch itself; or - a property of the cell of origin that was always present and only became visible once the lymphoid mask came off.
All three predict the same cross-sectional data. Paired diagnosis-to-switch sampling would separate them and is the obvious next study.
What it changes about the three immunotherapy failures
The graph's account of the CD19 failures is antigen-centric: lose CD19, lose the therapy. This adds a second clause with a different remedy.
- If the failure is antigen loss, the answer is a second antigen - CD22, CD33/CD123, CLEC12A, CD133 - and this graph already carries all of those.
- If the failure is also effector suppression, a second antigen does not help, because the T cell it recruits is being switched off. That argues for MDSC-directed strategies, for NK or iNKT effectors that are less MDSC-sensitive, or for pre-empting the switch rather than chasing it.
This graph's target-cd19-car-secreting-a-clec12a-bridge-against-lineage-switch is a pure antigen answer. The M-MDSC finding says an antigen answer may be necessary and not sufficient, and that is a testable prediction: a dual-antigen CAR should underperform in switched disease relative to its performance against a lineage-consistent myeloid leukaemia expressing the same antigen at the same density.
The measurement this disease still lacks
M-MDSC is a phenotypic and functional definition, not a genetic one: a monocytic surface phenotype plus demonstrated suppression of T-cell proliferation or function. Mikami supplies both halves 1, which is more than most MDSC papers do.
What is still missing in this disease: the immunoregulatory genes are 'distinct' but this beat could not extract the list from the abstract, so the graph cannot name the effector molecules - arginase, iNOS, IDO, PD-L1, reactive oxygen species are the classical candidates and none is confirmed here. That matters because each has a different drug. Reading the full text is the first thing a follow-up should do, and it is open access (PMC12381044).
| comparator | lineage-consistent KMT2A-rearranged AML (not healthy marrow) - which is what makes this a claim about the switch |
|---|---|
| not named in the abstract | the immunoregulatory effector molecules (arginase, iNOS, IDO, PD-L1, ROS) - each implies a different drug; full text open access at PMC12381044 |
| testable prediction | a dual-antigen CAR should underperform against switched disease relative to a lineage-consistent myeloid leukaemia matched for antigen density |
methods
[ "multi-omics of patient samples", "single-cell mass cytometry", "single-cell transcriptomics", "in vitro T-cell suppression assay" ]
two layers of escape
[ "antigen loss (CD19-negative myeloid phenotype)", "effector suppression (M-MDSC-like activity against T cells)" ]
unresolved causal alternatives
[ "installed by the transdifferentiation programme", "selected by CD19-directed pressure from a pre-existing suppressive subclone", "a property of the cell of origin unmasked by the switch" ]
- Adding a second antigen to a CAR does not address an effector that is being actively suppressed, so the graph's existing bridging and dual-target strategies may be necessary without being sufficient.
- Any menin-inhibitor immunology strategy delivered at relapse is delivered into this state, and menin inhibitors drive myelomonocytic differentiation - which is the direction of the M-MDSC phenotype, not away from it.
- If the suppressive population is selected rather than installed, then pre-emptive therapy intensification selects for it harder.
- MDSC-directed agents are broadly immunosuppressive or myelotoxic in their own right, and the patients in whom they would be used are heavily pre-treated infants with marrow failure.
- Is the M-MDSC-like state present in the diagnostic sample before any CD19-directed therapy? Paired diagnosis-to-switch sampling is the experiment, and the archived diagnostic material exists in the same cohorts.
- Which immunoregulatory effectors are responsible? Arginase, iNOS, IDO and PD-L1 each have a drug, and the abstract names none of them - a full-text read of PMC12381044 is the cheapest next step in this beat.
- Does the suppressive phenotype explain why lineage-switched relapse carries remission rates under 40%, or is that entirely accounted for by the change of target?
- Would an NK or iNKT effector be less affected? MDSC suppression is best characterised against T cells, and the CAR-iNKT platform in this disease would be a natural test.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Multi-omics analysis of patient samples found that lineage-switched acute myeloid leukaemia cells with KMT2A rearrangement possess monocytic myeloid-derived suppressor cell-like characteristics; single-cell mass cytometry showed an increase in the M-MDSC-like population in lineage-switched AML compared with lineage-consistent KMT2A-rearranged AML; single-cell transcriptomics identified distinct expression patterns of immunoregulatory genes within that population; and in vitro assays confirmed the immunosuppressive capacity of lineage-switched AML cells against T cells, analogous to that of myeloid-derived suppressor 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 2025What this evidence cannot show
Patient samples with no stated cohort size in the abstract, so the number of lineage-switched cases - an inherently rare event - is unknown and is likely small. Cross-sectional: samples are taken after the switch, so the design cannot distinguish a state installed by the switch from a pre-existing subclone selected by therapy. The immunoregulatory genes are described as distinct without being named in the abstract, so no effector molecule is identified here. In vitro suppression assays, no in vivo arm. Read: abstract only; record and identifiers fetched 2026-09-20; full text is open access at PMC12381044 and was not parsed by this beat.
Evidence that argues against this (1)
- A suppressive phenotype measured after a lineage switch is confounded with the intensive prior therapy that caused the switch: patients reaching lineage-switched relapse have had CD19-directed immunotherapy and usually much more, and therapy-associated myeloid suppressor expansion is well described independently of any leukaemia biology. The lineage-consistent comparator controls for genotype and lineage but not necessarily for treatment history. Mikami T et al. 2025
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Lineage switch to a CD19-negative myeloid phenotype is an established route of immune escape from CD19 CAR T-cell therapy in MLL-rearranged B-ALL.
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 2016What this evidence cannot show
Cited to establish the antigen-escape layer this entry adds a second layer to; the full treatment of lineage switch belongs to `risk-lineage-switch-cd19-directed-therapy` and `mechanism-lineage-switch-kmt2a` and is not duplicated here. Read: PubMed record and identifiers fetched 2026-09-20; content statement matches this graph's existing entries, with identifiers re-fetched rather than copied.
Evidence that argues against this (1)
- Frontline blinatumomab added to Interfant-06 chemotherapy in 30 KMT2A-rearranged infants produced a 4-year disease-free survival of 83.3% against 44.0% in controls with no lineage switch reported, so the escape route is real but does not outweigh the benefit of CD19 targeting in this population, and an immunosuppression argument must not be read as an argument against CD19-directed therapy. van der Sluis IM et al. 2023
- part of →Lineage switch in KMT2A-rearranged leukemia — the same clone changes what kind of blood cell it is
a second layer of escape beyond antigen loss: the switched cells suppress T cells
immunosuppression as well as antigen loss after CD19-directed therapy
- confers resistance to →A CD19 CAR that secretes a CLEC12A bridge: aimed squarely at lineage switch, and it inherits both antigens' failure modes
a second antigen does not help an effector that is being actively suppressed
route 5 of six documented in this disease