Oncofusion

start / graph / concept-mrd-monitoring-genomic-breakpoint

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

MRD by genomic breakpoint: the only assay in this disease that is fusion-specific by construction

also called genomic breakpoint MRD · KMT2A genomic fusion sequence MRD · patient-specific DNA junction PCR · DNA-based MRD · breakpoint qPCR

Every child's translocation joins the two genes at a slightly different DNA letter, so the join itself is a sequence that exists in that child's leukaemia cells and nowhere else. A PCR test built around that one join can find a single leukaemic cell among tens of thousands of normal ones, which is how doctors decide whether treatment is working.

The same thing, for a scientist

Genomic-breakpoint MRD uses the patient's own KMT2A-partner DNA fusion junction as the qPCR target. The junction is created by end joining with idiosyncratic microhomology and filler bases, so it is unique to the leukaemic clone; it sits on DNA rather than RNA, it is mono-allelic and present in every tumour cell regardless of transcription, and no physiological process can generate it. Diagnostic junction sequences are obtained from genomic DNA by long-distance inverse PCR, long-distance multiplex PCR or targeted NGS of the whole KMT2A gene; across 3,401 patients at one diagnostic centre at least one patient-specific fusion sequence was recovered in every case, with a roughly 95% discovery rate for the LD-PCR technologies and near 100% for targeted NGS [E1]. In Interfant-06 this was the **preferred** MRD-PCR target for KMT2A-rearranged infants, with Ig/TR rearrangements used as additional targets, and both target types achieved a quantitative range of at least 5x10^-4 in every patient [E2]. The choice was not cosmetic: once a cut-off was imposed, the KMT2A target predicted relapse better. 19 patients were positive above the cut-offs by KMT2A and not by Ig/TR (9 of 19 relapsed, 6-year cumulative incidence of relapse 48.7%), against 12 in the reverse direction (1 of 12 relapsed, CIR 8.3%, P = .02) [E2]. The 2007 European guideline framework for interpreting real-time quantitative PCR MRD data, written for Ig/TCR targets, supplies the quantitative-range and positivity rules these assays are read under [E3].

Go deeper
Why the junction is unique, and what follows from that

A KMT2A translocation breakpoint is repaired by end joining, and the repair leaves a junction whose exact base composition - including deletions, duplications and non-templated filler - differs between patients even when both break in the same intron and join the same partner exon. That has four consequences no other routine assay in this disease shares.

  1. Specificity is structural, not empirical. A primer pair spanning the junction has no template in any normal cell, so there is no biological background to subtract. Contrast Ig/TCR targets, where regenerating marrow contains physiological rearrangements and false positivity has been described 2.
  2. It is a DNA marker, so transcription does not matter. The recombinome authors state the point directly: a given KMT2A fusion allele is genetically stable and a mono-allelic marker for each tumour cell 1. A blast that has reduced or silenced fusion transcription still carries the junction.
  3. It needs the patient's own sequence. There is no off-the-shelf reagent. The diagnostic marrow must be captured and sent for breakpoint characterisation before the assay can exist, which is why concept-blood-film-and-blood-count-leukemia and concept-molecular-confirmation-fusion-transcript treat the diagnostic sample as the one chance to establish the target.
  4. Discovery is partner-agnostic. LDI-PCR and KMT2A-targeted NGS work outward from KMT2A and do not require the partner to be guessed 14, which matters in a disease with 107 characterised in-frame partners 1.
Sensitivity, quantitative range, and what the numbers actually mean

The clinically relevant figure is not a single sensitivity but the quantitative range - the lowest dilution at which the assay still gives a reproducible, quantifiable value. In Interfant-06, every patient's MRD-PCR target, whether KMT2A or Ig/TR, reached a quantitative range of at least 5x10^-4 2. That is the number the protocol's high, intermediate and negative cut-offs were built on. It is not a claim that the assay stops there, and many individual assays quantify to 10^-4 or below; it is the floor the trial could guarantee across every patient and every national reference laboratory.

The EuroMRD (ESG-MRD-ALL) framework exists precisely because MRD positive is a laboratory judgement rather than a reading 3. Thirty laboratories agreed rules for what counts as positive, what counts as positive-below-quantitative-range and what counts as negative, so that the same marrow gives the same risk-group assignment in Utrecht and in Monza. Two things follow: an MRD value quoted without its quantitative range is uninterpretable, and an MRD value from an unstandardised laboratory is not comparable with a trial's cut-point. That is dealt with separately in concept-what-mrd-negative-actually-means.

Turnaround, and the workflow that produces the target

The assay has two clocks. Establishing the target is slow: diagnostic genomic DNA goes to a breakpoint-characterisation centre, is digested and re-ligated for LDI-PCR or run through KMT2A-targeted NGS, the amplimer is sequenced, and the junction is returned to the sender for patient-specific assay design 1. Patient-specific primers are then designed and tested for sensitivity and specificity at the national reference laboratory before the first follow-up sample can be read 2. Running the assay once it exists is fast - a same-week qPCR on a marrow aspirate.

The practical consequence is that the first usable MRD timepoint in infant ALL is the end of induction, roughly five weeks in, and the assay must be commissioned on day one to be ready for it. A diagnostic pathway that stops at FISH has no target and no way to build one retrospectively without stored diagnostic material - the point made from the other side in concept-fish-split-signal-kmt2a.

Where this has been used outside infant ALL, and what it revealed

The same principle has been validated in other fusion leukaemias, and those comparisons are informative because they show what a fusion-specific DNA target detects that an Ig/TCR target does not. In childhood BCR::ABL1 ALL, DNA-based monitoring of the BCR-ABL1 genomic breakpoint was discordant with Ig/TCR in 32% of samples across 40 patients and 346 samples; the excess fusion signal marked multilineage involvement of the fusion-positive clone, with 12-83% of non-ALL B lymphocytes, T cells and/or myeloid cells carrying the fusion in discrepant patients - a CML-like biology 5. In ABL-class fusion ALL the two target types agreed closely (r^2 = 0.9786) 6. In two adult KMT2A-rearranged AML cases monitored through haploidentical transplantation a breakpoint-specific assay documented persistent molecular remission 7, and in an infant biphenotypic leukaemia with a cryptic MLLT10 insertion a breakpoint-derived assay rose before haematological relapse 8.

The BCR::ABL1 result is the cautionary one for this graph. A fusion-specific DNA target reports every cell carrying the rearrangement, including cells that are not the leukaemia being treated. In a disease whose founding lesion arises in a pre-VDJ stem or progenitor cell, that is not a hypothetical concern.

Facts and figures
assay classpatient-specific genomic DNA junction real-time quantitative PCR
analytegenomic DNA
targetKMT2A-partner der(11) genomic fusion junction, patient-specific
quantitative range achieved interfant06at least 5e-4 in all patients, both KMT2A and IG/TR targets
interfant06 rolemain MRD-PCR target; IG/TR used as additional targets
concordance with ig tcr interfant06
concordant_pct94.0concordant_pct: 94tp5_pct95.0tp5_pct: 95eoc_pct96.0eoc_pct: 96samples223.0samples: 223
discordance outcome interfant06
p_value0.020p_value: 0.02igtcr_only_relapses1.000igtcr_only_relapses: 1igtcr_only_cir_6y_pct8.300igtcr_only_cir_6y_pct: 8.3kmt2a_only_relapses9.000kmt2a_only_relapses: 9igtcr_positive_only_n12.0igtcr_positive_only_n: 12kmt2a_positive_only_n19.0kmt2a_positive_only_n: 19kmt2a_only_cir_6y_pct48.7kmt2a_only_cir_6y_pct: 48.7
target discovery methods
[
  "long-distance inverse PCR",
  "long-distance multiplex PCR",
  "targeted NGS of full-length KMT2A",
  "RNA-Seq (3 of 3401 cases)",
  "RT-PCR (1 of 3401 cases)"
]
target discovery rate
{
  "ld_pcr_pct": 95,
  "targeted_ngs_pct": 100,
  "source_pmid": "37019990"
}
concordance with ig tcr interfant06 — the numbers behind the chart
{
  "samples": 223,
  "concordant_pct": 94,
  "eoc_pct": 96,
  "tp5_pct": 95
}
discordance outcome interfant06 — the numbers behind the chart
{
  "kmt2a_positive_only_n": 19,
  "kmt2a_only_relapses": 9,
  "kmt2a_only_cir_6y_pct": 48.7,
  "igtcr_positive_only_n": 12,
  "igtcr_only_relapses": 1,
  "igtcr_only_cir_6y_pct": 8.3,
  "p_value": 0.02
}
cannot detect
[
  "KMT2A partial tandem duplication (no partner junction exists)",
  "KMT2A* (5'-truncated product of the gene-internal promoter; no junction exists)",
  "a subclone whose junction differs from the diagnostic one",
  "phenotype: the junction persists through lineage switch, so the assay cannot report it"
]
hands off from
[
  "concept-molecular-confirmation-fusion-transcript",
  "concept-fish-split-signal-kmt2a",
  "concept-blood-film-and-blood-count-leukemia"
]
related assays
[
  "concept-mrd-monitoring-fusion-transcript-rt-pcr",
  "concept-mrd-monitoring-ig-tcr-rearrangement",
  "concept-mrd-flow-cytometry-and-the-lineage-switch-blind-spot"
]
How this could bite us back
  • MRD drives intensification, and this assay's advantage over Ig/TCR appears only after a cut-off is imposed. The 19 patients positive by KMT2A and not by Ig/TR would be intensified on this assay and not on the other; 9 of them relapsed, but 10 did not, and those 10 would receive myeloid-style consolidation or transplant on the strength of a target-choice decision.
  • The assay is genetically stable by design, which means it reports the leukaemic genome and not the leukaemic phenotype. Under a differentiating agent, or after a lineage switch, the junction is still there, so a persistently positive breakpoint MRD does not distinguish resistant disease from a differentiated or switched clone that a different therapy should be aimed at.
  • It cannot see KMT2A*, the 5'-truncated species driven by the gene-internal promoter upstream of KMT2A exon 12, because there is no junction in it - and the recombinome authors note that the majority of reciprocal out-of-frame fusions may still express it. A treatment decision made on a negative breakpoint MRD is silent about that species.
  • Establishing the target consumes diagnostic material and specialist capacity. In a health system that cannot send DNA to a breakpoint-characterisation centre the recommended assay does not exist, and the child is monitored on the target the trial found inferior. That is an equity failure hiding inside a technical recommendation.
  • A fusion-specific DNA target will detect any persisting ancestral or pre-leukaemic clone carrying the same junction. In a disease that arises in utero in a pre-VDJ progenitor, persistent low-level positivity may not be residual leukaemia, and intensifying therapy against it would be treating a marker.
Still open
  1. What is the actual analytical sensitivity distribution of KMT2A breakpoint assays in infants? Interfant-06 reports only that every assay reached a quantitative range of at least 5x10^-4; the distribution below that line has never been published, and the difference between 5x10^-4 and 10^-5 is the difference between two very different clinical claims.
  2. Does the BCR::ABL1 multilineage finding have a KMT2A counterpart? If the fusion-positive clone in infant KMT2A-rearranged ALL extends beyond the B-lineage blast compartment - which the pre-VDJ cell-of-origin evidence would predict - a DNA junction assay would report cells that lymphoid-directed therapy is not treating, and the discordance would be biology rather than noise. Nobody has sorted a follow-up marrow in this disease and asked.
  3. How often is no breakpoint target obtainable in routine infant practice, as opposed to in the specialist referral network that developed the methods?
  4. Does a junction assay built on the diagnostic der(11) sequence stay valid at relapse in cases with complex or evolving rearrangements, and how often has a relapse been missed because the clone's junction changed?
Every source behind this entry (8)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Across 3,401 acute leukaemia patients, genomic KMT2A breakpoints were characterised by long-distance inverse PCR, long-distance multiplex PCR or targeted next-generation sequencing of the full-length KMT2A gene, and at least one patient-specific KMT2A fusion sequence was obtained in every case and used to establish patient-specific qPCR assays for MRD monitoring; the LD-PCR technologies had a discovery rate of about 95% and the KMT2A-targeting NGS method nearly 100%, and the authors state that a given KMT2A fusion allele is genetically stable and a mono-allelic marker for each tumour cell.

    The KMT2A recombinome of acute leukemias in 2023. · Meyer C, Larghero P, Almeida Lopes B, et al. · Leukemia 2023
    PMID 37019990DOI
    What this evidence cannot show

    Read in full text. This is a diagnostic-network case series, not a prospective MRD trial; the paper reports that patient-specific sequences were made available for MRD surveillance and cites prospective studies for the outcome evidence rather than reporting MRD performance itself. The cohort is referral-based (samples pre-screened elsewhere and sent to one centre), so partner and breakpoint frequencies are not population estimates. The 95% and 100% discovery rates are the centre's own and are not independently audited.

    Evidence that argues against this (2)
    • The same paper is explicit that a genomic analysis cannot provide functional information for out-of-frame fusions and that RNA-Seq is needed to see alternative splice events, so a DNA-only workflow characterises the MRD target while remaining blind to what the allele actually expresses. Meyer C et al. 2023
    • Targeted RNA-side NGS reaches the same partner-agnostic property from the transcript and detected KMT2A-AFF1 in MV4-11 dilutions to 10^-7 - a lower detection limit than any breakpoint qPCR reports - while failing to detect a KMT2A-MLLT3 fusion because transcript expression was low. Neither route is uniformly superior. Afrin S et al. 2018
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In Interfant-06 the KMT2A genomic breakpoint fusion sequence was the main MRD-PCR target for KMT2A-rearranged infants with immunoglobulin and T-cell receptor rearrangements as additional targets; all patients' targets had a quantitative range of at least 5x10^-4; results from the two target types were concordant in 94% of 223 samples; and applying protocol cut-offs, 19 patients were positive by KMT2A but not by IG/TR (9 relapses, 6-year cumulative incidence of relapse 48.7%) versus 12 in the reverse direction (1 relapse, CIR 8.3%, P = .02), leading the authors to conclude that the KMT2A genomic breakpoint fusion sequence is the preferred MRD target in this disease.

    PMID 33405950DOI
    What this evidence cannot show

    Read in full text. MRD analysis was not mandatory in Interfant-06 except for medium-risk patients at TP5, so the 249 analysed patients are a subset of 439 eligible; the authors report no significant difference in known prognostic features between those with and without MRD data, which bounds but does not remove selection. The 19-versus-12 comparison is a post-hoc subgroup analysis on small numbers with wide intervals (48.7%, 95% CI 23.8-69.7 versus 8.3%, 95% CI 0.4-32.6). Concordance was assessed only in samples where both targets existed, which excludes by construction the patients in whom no IG/TR target could be found.

    Evidence that argues against this (2)
    • The same analysis found no significant overall discrepancy between the two targets (P = .60 at end of induction; only 7 of 90 samples differed by at least one log, and 5 of those 7 were higher by KMT2A). On the primary agreement analysis the two targets are interchangeable, and the KMT2A advantage appears only after a cut-off is imposed. Stutterheim J et al. 2021
    • In KMT2A-germline infant ALL on the same protocol, where no KMT2A target exists and MRD must be IG/TR-based, MRD at end of induction was still prognostic (6-year DFS 61.4% for high MRD versus 87.9% for undetectable), so IG/TR-based MRD is not uninformative in infants as a class. Stutterheim J et al. 2022
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    The European Study Group on MRD detection in ALL, comprising 30 MRD-PCR laboratories worldwide, developed guidelines for interpreting real-time quantitative PCR MRD data so that the same data are interpreted identically between laboratories running one protocol and so that MRD results can be compared across treatment protocols.

    PMID 17287850DOI
    What this evidence cannot show

    The guideline was written for Ig/TCR targets; genomic fusion-breakpoint targets are read under the same quantitative-range and positivity framework by convention rather than because the guideline addresses them. It is a consensus document rather than a measurement, and the 2007 version predates droplet digital PCR and sequencing-based MRD.

    Evidence that argues against this (1)
    • Standardisation is not adoption. A ddPCR quality-control study notes that the existing real-time qPCR guidelines needed adaptation to cover droplet generation and intra- and inter-assay variation in the absence of standard curves, so the framework does not transfer unchanged to newer platforms. Chen D et al. 2021
  4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    The MLL recombinome series demonstrates that establishing patient-specific chromosomal fusion sites allows the design of specific PCR primers for minimal residual disease analysis for all patients, and documents an age-dependent breakpoint shift with intron 11 breakpoints associated with ALL and younger patients and intron 9 breakpoints predominating in AML or older patients.

    The MLL recombinome of acute leukemias in 2017. · Meyer C, Burmeister T, Gröger D, et al. · Leukemia 2018
    PMID 28701730DOI
    What this evidence cannot show

    The same referral-based diagnostic network and largely the same patient pipeline as E1, so E1 and E4 are not independent observations. The claim that primers can be designed for all patients is a statement about the 2,345 patients in whom a breakpoint was successfully characterised, not about all comers.

    Evidence that argues against this (1)
    • A targeted-NGS comparison from a different group notes that LDI-PCR, the gold standard for characterising MLL breakpoints, is laborious and requires a large input of genomic DNA - a real constraint in an infant marrow and one reason the assay is not universally available. Afrin S et al. 2018
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    DNA-based monitoring of the BCR-ABL1 genomic breakpoint in 48 children with ALL was discordant with Ig/TCR-based MRD in a substantial minority, with more than 20% of children having at least one log higher BCR-ABL1 than Ig/TCR or IKZF1-deletion levels, and cell sorting showed that 12% to 83% of non-ALL B lymphocytes, T cells and/or myeloid cells carried the fusion in the discrepant patients.

    PMID 28331056DOI
    What this evidence cannot show

    A different fusion and a different disease; BCR::ABL1 ALL has an established multilineage CML-like subset with no demonstrated counterpart in KMT2A-rearranged infant ALL. Cell sorting was performed on diagnostic material, so the inference about what the discordant follow-up signal represents is indirect.

    Evidence that argues against this (1)
    • In the same group's ABL-class fusion cohort, fusion-based and Ig/TCR MRD agreed closely (21 patients, 257 samples, r^2 = 0.9786), so multilineage discordance is a property of particular fusions rather than of DNA fusion targets in general. Venn NC et al. 2022
  6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Quantitative PCR assays designed from precise genomic breakpoints for ABL-class fusions and BCR-ABL1 were feasible in all patients tested and, for ABL-class fusions, gave high concordance with Ig/TCR MRD across 21 patients and 257 samples, with all MRD testing complying with EuroMRD guidelines.

    PMID 35650277DOI
    What this evidence cannot show

    ABL-class fusion ALL rather than KMT2A-rearranged disease; small patient numbers per fusion; retrospective re-measurement of previously tested remission samples rather than prospective parallel testing.

    Evidence that argues against this (1)
    • The same paper found Ig/TCR and BCR-ABL1 tests discordant in 32% of samples in BCR-ABL1 patients (r^2 = 0.4703), which shows the concordance result does not generalise across fusions even within one study. Venn NC et al. 2022
  7. E7weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

    A PCR method based on quantitative detection of the individual chromosomal breakpoint in the MLL gene documented complete and persistent molecular remission after haploidentical transplantation in two patients with MLL-rearranged AML who had relapsed after conventional chemotherapy.

    PMID 23211007DOI
    What this evidence cannot show

    Two patients, adults with AML rather than infants with ALL, and a case report rather than a series. It demonstrates feasibility in the transplant setting and nothing about sensitivity or predictive value.

    Evidence that argues against this (1)
    • A two-patient report showing molecular remission cannot separate assay performance from patient selection; the outcome evidence for breakpoint MRD rests on the Interfant cohorts rather than on case series. Stutterheim J et al. 2021
  8. E8weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

    In an infant with acute biphenotypic leukaemia carrying a cryptic MLLT10 insertion, long-distance inverse PCR on genomic DNA defined the fusion point and a real-time quantitative PCR using primers recognising the two fusion sites tracked MRD in parallel with other clinical markers, with the MRD level rising before haematological relapse.

    PMID 19380030DOI
    What this evidence cannot show

    Single case report. A rise-before-relapse observation in one patient carries no information about lead time or positive predictive value.

    Evidence that argues against this (1)
    • In the Interfant-06 cohort roughly a third of patients who were MRD-negative at end of consolidation still failed (6-year DFS 68.2%), so a breakpoint assay is not a reliable individual-level early-warning system. Stutterheim J et al. 2021
Connected to