Oncofusion

start / graph / concept-mrd-monitoring-fusion-transcript-rt-pcr

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

MRD by fusion transcript: fast, standardised, partner-limited - and structurally blind to KMT2A*

also called fusion gene transcript MRD · FGT-MRD · RT-qPCR MRD · RT-PCR MRD · fusion transcript monitoring

Instead of tracking the leukaemia's DNA join, this test tracks the RNA copy the join produces, which is present in many copies per cell and so is easy to detect. The catch is that it only works for fusions somebody has already designed primers for, and it measures how much RNA the cell is making rather than how many cells there are.

The same thing, for a scientist

Fusion-transcript MRD amplifies the chimeric mRNA across the exon-exon fusion junction by reverse-transcription quantitative PCR. It has been standardised since BIOMED-1, which built external, nested and independently shifted primer sets for nine chromosome aberrations including t(4;11) with MLL-AF4 [E1], and evidence-based RT-PCR designs now exist for the eight commonest KMT2A aberrations, together covering roughly 90% of KMT2A-rearranged cases [E2]. In infants it has been benchmarked head-to-head against flow cytometry in 942 bone marrow samples from 123 KMT2A-rearranged infants: overall concordance 87.4%, with fusion-transcript MRD the more reliable **late** relapse predictor while flow identified rapid early response [E3]. Its three structural limits are all consequences of using RNA as the analyte. First, signal is transcription-dependent, so a copy number is not a cell count, and a targeted RNA NGS assay missed a KMT2A-MLLT3 fusion outright because transcript expression was low [E4]. Second, it is partner-limited: a panel returns negative for every partner it was not designed around, against a recombinome of 107 characterised in-frame partners [E5]. Third, and specific to this gene, **KMT2A\* has no junction**: the 5'-truncated species driven by the gene-internal promoter that coincides with breakpoint cluster region hot spot II contains no partner sequence, so no junction primer pair can amplify it [E6]. Junction RT-PCR is structurally blind to it, not merely insensitive.

Go deeper
What has actually been standardised

The BIOMED-1 Concerted Action, 14 laboratories in eight European countries, selected nine well-defined chromosome aberrations with fusion gene transcripts and designed primers to predefined criteria for single PCR (external primers A-B) and nested PCR (internal primers C-D), with a designed confirmation step 1. t(4;11) with MLL-AF4 was one of the nine. That is why fusion-transcript MRD in this disease has a longer standardisation history than any other molecular assay here.

The modern equivalent is the set of evidence-based RT-PCR methods derived from more than 1,600 characterised genomic breakpoints, covering MLL-AF4, MLL-AF6, MLL-AF9, MLL-AF10, MLL-ENL, MLL-ELL, MLL-EPS15 and MLL partial tandem duplication - around 90% of KMT2A-rearranged cases 2. The design logic is worth noting: the primers were chosen from an empirical catalogue of which fusion transcript variants are actually observed, not from the reference transcript. The authors state that before this, no truly evidence-based RT-PCR methods had been published for most of these aberrations 2.

Head-to-head against flow in 942 infant marrow samples

The largest direct comparison in this disease studied 942 bone marrow samples from 123 KMT2A-rearranged infants by both multicolour flow cytometry and PCR for fusion gene transcripts 3. 383 samples (40.7%) were concordantly negative and 441 (46.8%) concordantly positive, giving 87.4% concordance. The disagreements were asymmetric: 99 samples (10.5%) were positive only by fusion transcript against 19 (2.0%) positive only by flow. Most discordance occurred near the sensitivity limits of the methods, and neither the type of KMT2A fusion nor a new treatment interfering with flow methodology changed the concordance rate.

The prognostic behaviour differed by timepoint, which is the more useful result: flow identified rapid response at early timepoints, fusion-transcript MRD was the reliable relapse predictor at later treatment stages, and the most precise risk definition came from combining the two 3. That is a direct argument against treating MRD platform choice as a single decision for the whole of therapy.

The three structural limits

RNA is not a cell count. Transcript copies per cell vary with fusion, with cell state and with treatment. The recombinome authors put the consequence plainly: RNA-based diagnostic techniques do not provide the patient-specific chromosomal fusion sequences used for MRD studies, and paired-end mRNA analysis does not exceed roughly 90% discovery because of variability in gene transcription 5. An anchored-multiplex NGS assay validated against LDI-PCR failed to detect a KMT2A-MLLT3 fusion for exactly this reason, while detecting KMT2A-AFF1 down to a 10^-7 dilution in the same study 4 - the dynamic range is excellent and the failure mode is categorical.

Partner-limited. A panel is negative for every partner it was not built around. With 107 characterised in-frame KMT2A fusions and 63 partners so far seen only once 5, a targeted panel is a prior, not a test. See the general form of this in concept-molecular-confirmation-fusion-transcript.

KMT2A* is invisible. The gene-internal promoter colocalising with breakpoint cluster region hot spot II generates an mRNA encoding an N-terminally truncated MLL that retains the C-terminal regions including the SET domain 6. There is no partner sequence in that transcript, so there is no junction to span. The recombinome notes that the majority of reciprocal out-of-frame KMT2A fusions may still be transcribed and encode this KMT2A* protein 5. A patient can therefore be fusion-transcript MRD-negative while a catalytically competent truncated KMT2A species is being made, and no routine assay counts it. concept-junction-rt-pcr-what-it-can-and-cannot-attribute develops the attribution side of this.

When the transcript assay is the only one available

In practice the fusion-transcript assay is the fallback where genomic breakpoint characterisation is not accessible - it needs only RNA, a thermal cycler and a published primer set, against a specialist LDI-PCR or targeted-NGS workflow at a referral centre. The MLL-Baby group's infant cohorts were monitored this way 3, and their conclusion was pragmatic: the two simple and inexpensive approaches of flow and fusion-transcript PCR are of high clinical value together.

That is a different claim from Interfant-06's, which found the genomic breakpoint the preferred target 7. The two are compatible. Interfant-06 compared genomic breakpoint against Ig/TR, not against fusion transcript; no trial has compared genomic-breakpoint MRD head-to-head against fusion-transcript MRD in infants. That comparison does not exist.

Facts and figures
assay classreverse-transcription quantitative PCR across the fusion exon-exon junction
analyteRNA
targetchimeric KMT2A-partner fusion transcript
measurestranscript abundance relative to a control gene, not cell number
comparison not doneno trial has compared genomic-breakpoint MRD against fusion-transcript MRD head-to-head in infant KMT2A-rearranged ALL
standardisation
[
  "BIOMED-1 Concerted Action (9 aberrations incl. t(4;11) MLL-AF4)",
  "evidence-based RT-PCR designs for the 8 commonest MLL aberrations, ~90% of KMT2A-r cases"
]
infant head to head vs flow
{
  "samples": 942,
  "patients": 123,
  "concordance_pct": 87.4,
  "fgt_only_positive_pct": 10.5,
  "flow_only_positive_pct": 2,
  "source_pmid": "34970734"
}
cannot detect
[
  "KMT2A* (5'-truncated gene-internal-promoter product: no junction exists)",
  "any partner the panel was not designed around",
  "a fusion whose transcript is expressed below assay threshold in that patient",
  "protein-level persistence (RNA is not protein)"
]
How this could bite us back
  • Fusion-transcript MRD is the cheap, widely deployable assay, so it is the one most likely to be used where resources are tight - and it is the one whose signal moves with transcription. Any therapy that changes fusion transcript abundance without killing cells will look like a response.
  • A negative result on a targeted panel is an absence of testing, not a negative result, for every partner outside the panel. In a disease with 63 partners seen once each, a child with a rare partner can be reported MRD-negative by an assay that was never capable of detecting their leukaemia.
  • The assay cannot see KMT2A*, which retains the SET domain and therefore catalytic competence. If the fusion's contribution is suppressed by a menin inhibitor while KMT2A* persists, a junction-based transcript readout would report molecular remission of the species that stopped mattering.
  • Combining flow and transcript MRD gave the best risk definition in the only large infant comparison, which means single-platform monitoring is systematically discarding information that is already being generated - and the discarded 10.5% of transcript-only-positive samples are the ones an intensification decision would turn on.
Still open
  1. How many fusion transcript copies per leukaemic cell are there in KMT2A-rearranged infant ALL, and does that number change under menin inhibition or under lineage switch? Without it, a transcript MRD value cannot be converted into a cell burden, and the two therapies most likely to change it are the two now entering the frontline.
  2. Is KMT2A* expressed in infant KMT2A-rearranged blasts at diagnosis and in remission marrow? A promoter-specific RT-qPCR with a forward primer in the gene-internal first exon would settle it and requires no new technology.
  3. Does a genomic-breakpoint assay and a fusion-transcript assay run on the same infant marrow samples agree? Interfant used one, MLL-Baby the other, and nobody has run both.
Every source behind this entry (7)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    The BIOMED-1 Concerted Action standardised RT-PCR analysis of fusion gene transcripts for MRD detection across 14 laboratories in eight European countries, selecting nine well-defined chromosome aberrations including t(4;11) with MLL-AF4, and designing primers to predefined criteria for single PCR (external primers A-B) and nested PCR (internal primers C-D).

    PMID 10602411DOI
    What this evidence cannot show

    A 1999 methodological standardisation report. It establishes that primers and a confirmation step were designed and quality-controlled, not that the assay predicts outcome, and predates every modern platform. The nine aberrations chosen reflect what was recognised in 1999; MLL-AF4 is the only KMT2A fusion in the set.

    Evidence that argues against this (1)
    • A 2015 analysis from a large breakpoint archive states that no truly evidence-based RT-PCR methods had been published for most common MLL aberrations before that point, which implies the 1999 standardisation did not in practice cover the KMT2A fusion landscape. Burmeister T et al. 2015
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Using a collection of more than 1,600 MLL genomic breakpoints in acute leukaemias at one diagnostic centre, evidence-based RT-PCR methods were developed for the eight most common MLL aberrations (MLL-AF4, MLL-AF6, MLL-AF9, MLL-AF10, MLL-ENL, MLL-ELL, MLL-EPS15 and MLL partial tandem duplication), together accounting for around 90% of MLL-rearranged cases, and the authors state that no truly evidence-based RT-PCR methods for most of these aberrations had been published before.

    PMID 25510485DOI
    What this evidence cannot show

    The primer designs were derived from the observed fusion transcript variants in one referral archive, so coverage of rarer variants is bounded by what that archive saw. The paper reports assay design and expected coverage rather than prospective clinical sensitivity or MRD performance, and the 90% figure describes the proportion of KMT2A-rearranged cases carrying one of the eight aberrations, not assay sensitivity within them.

    Evidence that argues against this (1)
    • The 2023 recombinome from the same network lists 107 characterised in-frame KMT2A fusions with 63 partners so far identified only once, so an eight-aberration panel leaves a long tail for which no validated RT-PCR design exists and a negative result carries no information. Meyer C et al. 2023
  3. E3strong 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, MRD by multicolour flow cytometry and by PCR for fusion gene transcripts were concordant in 87.4% of samples (383 concordantly negative, 441 concordantly positive), with 99 samples (10.5%) positive only by fusion transcript and 19 (2.0%) only by flow; most discordance occurred near the sensitivity limits; flow identified rapid response at early timepoints while fusion-transcript MRD was a reliable relapse predictor at later treatment stages, and combining the two gave the most precise risk definition.

    PMID 34970734DOI
    What this evidence cannot show

    Single-group cohort (MLL-Baby protocol) treated on a moderate-intensity regimen that differs from Interfant, so the timepoints and the treatment pressure at each are not transferable without care. Concordance is a sample-level statistic and does not by itself establish which method should drive a decision. The paper does not report the quantitative range of either assay per patient.

    Evidence that argues against this (1)
    • Interfant-06 found the genomic KMT2A breakpoint the preferred MRD target and did not test fusion-transcript MRD at all, so the two protocols recommend different molecular targets from non-overlapping comparisons. Stutterheim J et al. 2021
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    An anchored multiplex PCR-based targeted next-generation sequencing assay validated against long-distance inverse PCR in 39 paediatric leukaemia specimens identified ten different MLL fusions without prior knowledge of the partner and detected MLL-AFF1 in MV4-11 cell dilutions of 10^-7 and transcripts down to 0.005 copies per nanogram, but failed to detect an MLL-MLLT3 fusion because of low expression of the fusion transcript.

    Targeted Next-Generation Sequencing for Detecting MLL Gene Fusions in Leukemia. · Afrin S, Zhang CRC, Meyer C, et al. · Mol Cancer Res 2018
    PMID 29133595DOI
    What this evidence cannot show

    39 diagnostic specimens, not remission samples: the 10^-7 figure is a cell-line dilution sensitivity, not a validated clinical MRD limit of detection in patient marrow. The single MLLT3 miss is n=1 and establishes the failure mode rather than its frequency.

    Evidence that argues against this (1)
    • In the same comparison the RNA-side assay outperformed LDI-PCR in one case, identifying an MLL-MLLT1 fusion where LDI-PCR failed to determine the breakpoint - so transcript-based detection is not uniformly the weaker route. Afrin S et al. 2018
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The 2023 KMT2A recombinome states that RNA-based diagnostic techniques do not provide the patient-specific chromosomal fusion sequences that may be used for MRD studies and do not allow paired-end mRNA analysis discovery rates greater than 90% because of variability in gene transcription and bioinformatic problems, and that the majority of reciprocal out-of-frame KMT2A fusions may still be transcribed and encode a 5'-truncated KMT2A protein termed KMT2A* from a gene-internal promoter upstream of KMT2A exon 12.

    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 methodological assertion by a group whose own platform is DNA-based, so the comparison is not disinterested; the 90% figure is stated rather than measured in a head-to-head. The KMT2A* statement is an inference from promoter position applied to the reciprocal out-of-frame set, not a measurement of KMT2A* protein in those patients.

    Evidence that argues against this (1)
    • The same paper concedes that RNA-Seq provides insights into alternative splice events that genomic analysis cannot supply, explicitly for the out-of-frame reciprocal cases - so the DNA-only position is stated with its own exception attached. Meyer C et al. 2023
  6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    A gene-internal promoter element in the human MLL gene colocalises with recombination hot spot II in the breakpoint cluster region, and the mRNA it generates encodes an N-terminally truncated version of MLL that still exhibits many functional regions including the C-terminal SET domain.

    PMID 16983345DOI
    What this evidence cannot show

    The promoter and the truncated transcript were characterised in cell systems and in mouse and human MLL, not quantified in patient leukaemia samples. The inference that this species is invisible to junction RT-PCR is a structural argument from its lack of partner sequence rather than an experimental demonstration of assay failure.

    Evidence that argues against this (1)
    • No published study has measured KMT2A* transcript or protein in a patient cohort under MRD monitoring, so its clinical relevance to an MRD-negative result is unestablished; the claim here is about what the assay can see, not about what is there. Meyer C et al. 2023
  7. E7established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Interfant-06 used the KMT2A genomic breakpoint fusion sequence as the main MRD-PCR target with immunoglobulin and T-cell receptor rearrangements as additional targets, and did not use fusion transcript RT-PCR as an MRD target.

    PMID 33405950DOI
    What this evidence cannot show

    Read in full text. This establishes what Interfant-06 did, and therefore that no head-to-head of genomic-breakpoint against fusion-transcript MRD exists inside that trial; it is not evidence that fusion-transcript MRD would have performed worse.

    Evidence that argues against this (1)
    • The MLL-Baby infant cohorts were monitored successfully on fusion-transcript PCR with flow, and their authors describe the two as simple, inexpensive approaches of high clinical value - so the Interfant choice reflects an available specialist network as much as an assay comparison. Popov A et al. 2023
Connected to