Oncofusion

start / graph / concept-mrd-flow-cytometry-and-the-lineage-switch-blind-spot

Explainerstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q34-mrd-monitoring

Flow-cytometric MRD: the fastest readout, and the one a lineage switch walks straight out of

also called MFC-MRD · flow MRD · multiparameter flow cytometry MRD · FCM-MRD · immunophenotypic MRD

Flow cytometry finds leftover leukaemia by looking for cells wearing the wrong combination of surface markers, and it gives an answer the same day. Its weakness in this disease is that the leukaemia can change which markers it wears - sometimes turning into a myeloid leukaemia entirely - and a panel built to find B-lineage cells will then report a clean marrow.

The same thing, for a scientist

Multiparameter flow cytometry MRD identifies residual blasts by a leukaemia-associated immunophenotype against the normal maturation pattern. It is the fastest and cheapest MRD method, needs no patient-specific reagent, and is applicable in essentially every patient. In infant KMT2A-rearranged ALL it works: on COG AALL0631, MRD by multiparameter flow at three timepoints separated outcomes in 117 KMT2A-rearranged infants (3-year EFS 25% versus 49% at TP1, 21% versus 47% at TP2, 22% versus 51% at TP3) while being non-prognostic in 58 KMT2A-germline infants [E1]. Against molecular MRD it is slightly less sensitive and differently timed: across 942 marrow samples from 123 KMT2A-rearranged infants, concordance with fusion-transcript PCR was 87.4%, with 10.5% of samples positive only by the molecular assay against 2.0% only by flow, flow identifying rapid early response and the molecular assay predicting relapse at later stages [E2]. Classical flow is generally held not to reach a solid 10^-4 [E3], and high-throughput sequencing of IGH and TRG identified 38.7% more MRD-positive patients than flow at a 0.01% threshold in 619 paired paediatric B-ALL samples, with the extra patients having worse outcomes than flow-negative ones [E4]. **The subtype-specific failure is lineage switch.** A KMT2A-rearranged clone can convert to a CD19-negative myeloid phenotype under CD19-directed pressure [E5]; a B-lineage-keyed MRD panel cannot follow it, and the switch has been documented from a state of MRD-negative remission of the B-ALL.

Go deeper
What flow MRD can do that the molecular assays cannot

Three things, and they are not trivial.

Speed and universality. No patient-specific reagent, no breakpoint characterisation, no reference laboratory. In a disease where the specialist molecular target takes weeks to commission, flow gives an answer on the diagnostic marrow and on every marrow thereafter. In KMT2A-germline infant ALL on the MLL-Baby protocol, where genetics gave no prognostic discrimination at all - not age, not the newly described NUTM1 and PAX5 fusions - only rapidity of response measured as flow MRD had a statistically significant impact on outcome 6.

Early timepoints. The infant head-to-head found flow able to identify rapid response at early timepoints where the molecular assay did not, and the molecular assay the better late relapse predictor; combining the two gave the most precise risk definition 2. That is an argument for running both, not for choosing.

It sees phenotype. This is the same property that makes it vulnerable, and it cuts both ways: flow is the only routine assay that would notice a myeloid population appearing - if the panel included myeloid markers and if the operator was looking for it.

The lineage-switch blind spot, stated precisely

The failure is not that flow is insensitive. It is that the quantity flow measures stops being the quantity of interest.

Seven children with KMT2A-rearranged B-ALL treated with CD19 CAR-T cells all achieved complete remission in the marrow by flow cytometry; within one month of infusion two developed AML that was clonally related to their B-ALL 5. The B-lineage readout was correct and irrelevant. This graph holds the same phenomenon under CD19-directed therapy as risk-lineage-switch-cd19-directed-therapy, under blinatumomab with spontaneous reversion when the drug was stopped 7, and under menin inhibition, where in a sample that lineage switched during revumenib treatment the oncoprotein and menin fell to undetectable levels while ENL persisted on target loci including genes in the granulocyte-monocyte-progenitor lineage-switching programme 8.

Three consequences for how an MRD result should be read in this disease:

  1. A negative B-lineage flow MRD under CD19-directed or menin-directed therapy is a conditional negative. It says the B-ALL is not detectable. It does not say the clone is gone.
  2. The molecular assays are the ones that survive a switch, because the genomic junction and, in a switched clone retaining expression, the fusion transcript are lineage-independent. The Wolfl case shows the phenotype reverting within days of stopping the drug 7 - a clone that was there throughout.
  3. Panel design is the mitigation and is not automatic. The switched population in the blinatumomab case remained at least partially positive for the NG2-directed antibody 7.1 7, and KMT2A-rearranged blasts characteristically co-express CD15, CD65 and CD33 at diagnosis, so a panel carrying myeloid markers and NG2 has a chance of following the clone. A panel keyed to CD19/CD10/CD34 does not.
Differentiation therapy is the second phenotype problem

Menin inhibition in myeloid disease produces differentiation, with differentiation syndrome as a recognised toxicity in the registrational trial 9. A flow MRD panel discriminates blasts from normal maturing cells by maturation pattern. A drug whose mechanism is to push leukaemic cells along a normal-looking maturation pathway therefore attacks the assay's discriminant directly: the cells become harder to call abnormal while remaining clonal.

The infant head-to-head reports that a new type of treatment hampering MFC methodology did not change the concordance rate in their cohort 2, which is reassuring but is one cohort and one treatment. This graph has already recorded that menin inhibition behaves as differentiation therapy in myeloid disease and as cytotoxic therapy in lymphoid disease, so the concern is sharpest in KMT2A-rearranged AML and in a lymphoid clone that has switched - which is precisely the population in which the phenotype is already unreliable.

No published study has measured flow MRD against a lineage-independent molecular MRD assay in patients on a menin inhibitor. That comparison does not exist, and it is the one that would settle whether menin-inhibitor response rates reported as MRD-negative remission are measuring cell clearance or cell maturation.

Facts and figures
assay classmultiparameter flow cytometry on fresh bone marrow or blood
analytesurface and cytoplasmic protein on intact cells
targetleukaemia-associated immunophenotype against normal maturation pattern
patient specific reagent requiredfalse
typical threshold0.01% (1e-4) in COG protocols
comparison not doneno published comparison of flow MRD against lineage-independent molecular MRD in patients receiving a menin inhibitor
infant kmt2a r prognostic
{
  "study": "COG AALL0631",
  "n": 117,
  "timepoints": 3,
  "efs_3y_pos_vs_neg_pct": [
    [
      25,
      49
    ],
    [
      21,
      47
    ],
    [
      22,
      51
    ]
  ],
  "source_pmid": "37259259"
}
concordance with fusion transcript
{
  "samples": 942,
  "patients": 123,
  "concordant_pct": 87.4,
  "molecular_only_positive_pct": 10.5,
  "flow_only_positive_pct": 2,
  "source_pmid": "34970734"
}
blind spots
[
  "lineage switch to a myeloid phenotype (panel is keyed to B-lineage)",
  "CD19 loss under CD19-directed therapy",
  "differentiation under a differentiating agent erodes the blast/normal discriminant",
  "sensitivity floor: classical flow not a solid 1e-4"
]
mitigations supported by evidence
[
  "include myeloid markers CD15/CD65/CD33 and NG2/CSPG4 in the panel",
  "pair with a lineage-independent molecular assay",
  "re-phenotype rather than re-run the old panel at any suspicion of switch"
]
How this could bite us back
  • A B-lineage flow MRD panel returning negative is the exact signal that precedes the worst outcome in this disease: all seven CAR-T-treated KMT2A-rearranged patients in the founding series were in flow-defined marrow complete remission, and two of them had AML within a month. Treating that negative as reassurance is the failure mode.
  • Flow MRD is what most of the world can afford. Saying it is the weaker assay without funding the alternative means the children monitored on it are told their result is less trustworthy and nothing changes.
  • Menin inhibition is being added to frontline infant therapy. If it differentiates cells in some patients, the assay that defines response in the registrational trials is the assay most disturbed by the mechanism, and the effect would make the drug look better rather than worse.
  • Combining flow with a molecular assay gave the best risk definition in the only large infant comparison. Any protocol that intensifies on a single platform is discarding information it already paid to generate, and the discarded samples are concentrated at the decision threshold.
  • Flow is the only routine assay that could detect a switch, and it detects one only if the panel was designed to. The mitigation for flow's blind spot is a change to flow, which means the assay's weakness is also the only place the fix can go.
Still open
  1. Under a menin inhibitor, does flow MRD diverge from a lineage-independent molecular MRD assay? Nobody has run both on the same marrows in a menin-inhibitor trial, and the answer decides whether reported MRD-negative remissions measure clearance or maturation.
  2. How often is a lineage switch detected only because someone re-phenotyped rather than re-ran the diagnostic panel? The published cases are those that were found; the denominator is unknowable from case reports and would need a protocol mandating a myeloid panel at every MRD timepoint in KMT2A-rearranged disease.
  3. Does adding NG2/CSPG4 to an MRD panel actually follow a switched clone? The one case with data remained at least partially 7.1-positive after switching, which is a single observation on which a panel-design recommendation is currently resting.
  4. Why was flow MRD prognostic in KMT2A-germline infants on MLL-Baby and not on AALL0631? The two results are in opposite directions in the same biological group and neither cohort is large enough to settle it.
Every source behind this entry (9)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In Children's Oncology Group study AALL0631, MRD measured by multiparameter flow cytometry at three timepoints separated outcomes in 117 infants with KMT2A-rearranged ALL (3-year event-free survival for MRD-positive at or above 0.01% versus MRD-negative: 25% versus 49% at TP1, 21% versus 47% at TP2, 22% versus 51% at TP3) but was not prognostic in 58 infants with KMT2A-germline ALL, and MRD was a strong independent outcome predictor in KMT2A-rearranged but not KMT2A-germline infant ALL.

    PMID 37259259DOI
    What this evidence cannot show

    The KMT2A-germline arm has only 58 patients with few events, so the failure to find prognostic value there is as compatible with low power as with a real absence. The 0.01% threshold is the standard COG flow cut-off and was not derived in infants. TP1-TP3 are protocol timepoints and are not directly comparable with the Interfant end-of-induction and end-of-consolidation timepoints.

    Evidence that argues against this (1)
    • On the Interfant-06 protocol, molecular MRD was prognostic in KMT2A-rearranged infants and MRD was also prognostic in the KMT2A-germline cohort on the same protocol (6-year DFS 61.4% for high end-of-induction MRD versus 87.9% undetectable), so the KMT2A-germline null in AALL0631 does not replicate across protocols and methods. Stutterheim J et al. 2022
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In 942 bone marrow samples from 123 infants with KMT2A-rearranged ALL, multicolour flow cytometry MRD and PCR for fusion gene transcripts were concordant in 87.4% of samples, with 99 samples (10.5%) positive only by fusion transcript against 19 (2.0%) only by flow; most discordance occurred near the sensitivity limits; neither the type of KMT2A fusion nor a new type of treatment hampering flow methodology influenced the concordance rate; flow identified rapid response at early timepoints while fusion-transcript MRD was a reliable relapse predictor at later stages; and the most precise risk definition came from combining the two.

    PMID 34970734DOI
    What this evidence cannot show

    Single-group MLL-Baby cohort on a moderate-intensity protocol; timepoints differ from Interfant. The statement that a new treatment hampering flow methodology did not affect concordance is reported without naming the treatment or quantifying the affected subset, so it cannot be read as a test of differentiation therapy.

    Evidence that argues against this (1)
    • The asymmetry of discordance (10.5% molecular-only versus 2.0% flow-only) means flow misses roughly five times as many positive samples as it uniquely finds, so a combined strategy is not symmetric and flow is the weaker single platform at the sample level. Popov A et al. 2023
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Flow cytometry and PCR analysis of rearranged Ig and TCR genes are the two established MRD techniques in ALL, but classical flow cytometry does not reach a solid 10^-4 while classical allele-specific oligonucleotide PCR is time-consuming and labour intensive, which is why high-throughput sequencing and next-generation multidimensional flow cytometry are being explored.

    PMID 25999452DOI
    What this evidence cannot show

    A review written by the developers of both the EuroFlow and EuroMRD standards, so it is an interested description of the field. The 10^-4 statement describes classical flow and is explicitly contrasted with next-generation flow, which the same authors developed; modern 8-10 colour panels with high cell acquisition perform better than the statement implies.

    Evidence that argues against this (1)
    • In the largest infant comparison, flow was uniquely positive in 2.0% of 942 samples and most discordance was at the sensitivity limits, so the practical sensitivity gap in this specific disease is smaller than a general statement about 10^-4 suggests. Popov A et al. 2023
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In 619 paired pretreatment and end-of-induction bone marrow samples from Children's Oncology Group studies AALL0331 and AALL0232, high-throughput sequencing of IGH and TRG identified 55 (38.7%) more patients as MRD-positive at a 0.01% threshold than flow cytometry did, and those discrepant patients had worse outcomes than flow-MRD-negative patients.

    PMID 29284596DOI
    What this evidence cannot show

    Paediatric B-ALL generally, not infants and not KMT2A-rearranged disease; AALL0331 was standard-risk and AALL0232 high-risk with day-29 flow MRD below 0.1%, so the cohort is selected by flow result in a way that affects the comparison. The result establishes that flow misses molecularly detectable disease, not that flow is wrong about the biology.

    Evidence that argues against this (1)
    • The same study reports that flow and sequencing gave similar 5-year EFS and OS for MRD-positive and MRD-negative patients at the 0.01% threshold, so on the primary prognostic comparison the two methods performed alike and the advantage is in the reclassified minority. Wood B et al. 2018
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Seven patients with B-cell acute lymphoblastic leukaemia harbouring MLL rearrangement were treated with CD19-specific CAR-T cells; all achieved complete remission in the bone marrow by flow cytometry, and within one month of CAR-T-cell infusion two developed acute myeloid leukaemia that was clonally related to their B-ALL, described by the authors as a novel mechanism of CD19-negative immune escape.

    PMID 26907630DOI
    What this evidence cannot show

    Seven patients, two events; a case series establishing that the failure mode exists, not its frequency. Marrow complete remission was assessed by flow cytometry, which is the assay whose blind spot the case illustrates - the clonal relationship was established by other means.

    Evidence that argues against this (1)
    • A single course of sequential blinatumomab added to Interfant-06 chemotherapy improved 2-year disease-free survival to 82% against 49% for historical chemotherapy alone, so CD19-directed therapy in this population delivers a large benefit that the switch risk does not outweigh, as recorded in risk-lineage-switch-cd19-directed-therapy. Ray A et al. 2025
  6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Among 39 infants with KMT2A-germline B-cell precursor ALL on the MLL-Baby protocol, none of the initial parameters including genetics and the presence of NUTM1 and PAX5 fusions distinguished patients with different outcomes, and only rapidity of response measured as minimal residual disease by flow cytometry showed a statistically significant impact.

    PMID 36715125DOI
    What this evidence cannot show

    39 patients with 5 relapses. With that many candidate baseline variables and that few events, a single significant factor is weak evidence, and the negative findings on genetics are underpowered rather than informative. KMT2A-germline disease, so it does not speak to the rearranged subtype this beat is about.

    Evidence that argues against this (1)
    • COG AALL0631 found flow MRD non-prognostic in 58 KMT2A-germline infants at all three timepoints, the opposite result in the same biological group, so the prognostic value of flow MRD in KMT2A-germline infant ALL is not settled. Faulk KE et al. 2023
  7. E7moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    An infant with KMT2A-rearranged ALL underwent a blinatumomab-induced ALL-to-AML lineage switch that spontaneously reverted, illustrating the high plasticity of this phenotype in MLL-rearranged leukaemia adapting rapidly to immunological stimuli.

    PMID 29898879DOI
    What this evidence cannot show

    A single case letter; the published abstract is one sentence and the numerical detail cited elsewhere in this graph comes from the body text. A reversion in one patient shows plasticity, not its frequency, and says nothing about how often a flow panel missed a switch that did not revert.

    Evidence that argues against this (1)
    • Most documented lineage switches do not revert and carry remission rates under 40% with median overall survival of 4.8 months, so a reverting case is the favourable extreme of the phenomenon rather than typical. Gardner R et al. 2016
  8. E8moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Profiling oncoprotein target sites in 36 MLL-rearranged leukaemia samples including three that underwent lymphoid-to-myeloid lineage switching in response to therapy showed that in lineage-switching samples oncoprotein levels are reduced and the oncoproteins preferentially activate granulocyte-monocyte progenitor genes, and that in a sample that lineage switched during treatment with the menin inhibitor revumenib the oncoprotein and menin were reduced to undetectable levels while ENL persisted on numerous oncoprotein-target loci including genes in the GMP-like lineage-switching programme.

    MLL oncoprotein levels influence leukemia lineage identities. · Janssens DH, Duran M, Otto DJ, et al. · Nat Commun 2024
    PMID 39472576DOI
    What this evidence cannot show

    The revumenib lineage-switch observation is n=1 within a 36-sample profiling study; three lineage-switching samples in total. This is chromatin profiling, not an MRD study, and it does not report what any MRD assay showed in that patient. The inference that flow MRD would have been misleading is ours.

    Evidence that argues against this (1)
    • In the same study, oncoprotein genomic enrichment was highly variable between samples and dynamically regulated, so a single switching sample under revumenib cannot be distinguished from the general variability the paper documents. Janssens DH et al. 2024
  9. E9strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In the phase II registrational portion of AUGMENT-101, 94 patients with relapsed or refractory KMT2A-rearranged acute leukaemia treated with revumenib had grade 3 or higher differentiation syndrome in 16.0%, and among responders 15 of 22 (68.2%) had no detectable residual disease.

    PMID 39121437DOI
    What this evidence cannot show

    The abstract does not state which MRD assay was used, at what sensitivity, or in what tissue, and the 15-of-22 denominator is a subset of responders rather than of treated patients. Differentiation syndrome incidence is a toxicity figure and is used here only to establish that differentiation occurs under this drug, not to quantify its effect on any assay.

    Evidence that argues against this (1)
    • The phase I report of the same drug states both that residual disease was cleared using sensitive clinical assays and that hallmarks of differentiation into normal haematopoietic cells were identified, including differentiation syndrome. The two observations are reported together and the paper does not separate them, so the confound is present in the founding evidence rather than introduced by later trials - and equally, differentiation being present does not show that the MRD clearance was artefactual. Issa GC et al. 2023
Connected to