Oncofusion

start / graph / concept-reciprocal-status-and-breakpoint-position-in-mrd-assay-design

Explainermoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q34-mrd-monitoring

Is reciprocal status or breakpoint position used in MRD design or risk stratification? Once, by necessity, in 16 of 3,401 patients - and nowhere in any risk model

also called reciprocal fusion MRD target · der(4) junction MRD · breakpoint position risk stratification · AF4-MLL prognostic · reciprocal allele assay design

The translocation makes two joined chromosomes, and the leftover-disease test is normally built on just one of them. It has been built on the other only in the rare patients where the usual one could not be found, and neither which chromosome is tracked nor exactly where the gene broke is used to decide how intensively any child is treated - even though one study found the second product strongly predicts survival.

The same thing, for a scientist

The answer in three parts. **(1) Reciprocal status enters assay design only as a fallback.** Across 3,401 patients, the diagnostic network's stated goal was *at least one* patient-specific KMT2A fusion sequence per patient for qPCR MRD, obtained by LDI-PCR, LD multiplex PCR or KMT2A-targeted NGS [E1]. In **16** patients the der(11) could not be identified but a reciprocal fusion allele was diagnosed, and in those the reciprocal junction is by construction the only available target [E1]. Two further patients had a 5'-KMT2A deletion with a reciprocal fusion present [E1]. So reciprocal status is used, in roughly 0.5% of cases, because there is nothing else. **(2) Breakpoint position is used to build the assay and is not used to stratify risk.** Intron position determines which restriction fragment LDI-PCR recovers and where the primers sit, and the recombinome documents a systematic age- and lineage-dependent breakpoint shift [E2]; but Interfant-06's risk model is KMT2A status, age, WBC and prednisone response [E3], and its multivariable model did not include breakpoint position at all. **(3) Reciprocal expression has been shown to be strongly prognostic and is still not used.** In 43 t(4;11) infants, AF4-MLL was expressed in only 45% (19 of 43); those patients had 4-year event-free survival 62.4% versus 11.7% and overall survival 73.7% versus 25.2%, and in a Cox model adjusting for Interfant-06 risk stratification, lacking AF4-MLL expression carried a hazard ratio of 3.42 (95% CI 1.35-8.63, P = 0.01) **while risk group itself was not significant** (HR 1.34, 95% CI 0.59-3.03, P = 0.49) [E4].

Go deeper
What the assay designers actually do with the reciprocal allele

The recombinome network's methods answer this directly. Genomic DNA is digested and re-ligated for LDI-PCR, or run through LD multiplex PCR or targeted NGS of full-length KMT2A; this identifies direct and reciprocal KMT2A fusions, complex rearrangements, insertions and deletions; and as a result, at least one patient-specific KMT2A fusion sequence was obtained and used for establishing patient-specific qPCR assays 1.

At least one. The workflow does not systematically build a two-target assay. Where both junctions are recovered, nothing in the published methods says the reciprocal is used, and the Interfant-06 MRD paper describes a single KMT2A target per patient alongside IG/TR 5. The only documented circumstance in which the reciprocal is the tracked target is when the der(11) junction cannot be found: 16 patients of 3,401 where no direct KMT2A fusion was identified but a reciprocal allele was, plus 2 with a 5'-KMT2A deletion and a reciprocal fusion, plus 3 of the 18 patients whose partner was an intergenic region 1.

This matters for a specific reason this graph already cares about. Roughly half of KMT2A::MLLT1 cases (241 of 460) break upstream of MLLT1 exon 1 in the spliced-fusion architecture and express no reciprocal transcript at all 1. In a patient with no reciprocal allele, a two-target assay is impossible; in a patient with one, nobody has published whether building it would help.

The prognostic finding nobody stratifies on

Reciprocal AF4-MLL expression is, on the published evidence, a stronger predictor in t(4;11) infants than the entire Interfant-06 risk model. In 43 t(4;11) patients uniformly diagnosed and treated on Interfant 99/06, AF4-MLL was expressed in 19 (45%), its expression correlated with HOXA cluster expression (t-test P = 0.002), and the AF4-MLL/HOXA-expressing group had 4-year EFS 62.4% (SE 11.3) against 11.7% (SE 10.2) and OS 73.7% against 25.2% 4. Adjusted for Interfant-06 risk group, the hazard ratio for lacking AF4-MLL expression was 3.42 while the risk group's own hazard ratio was 1.34 and non-significant 4.

Four reasons this has not entered practice, only the first of which is about the evidence:

  1. n = 43, single study, not replicated. Wide standard errors on both survival estimates and a Cox model on a few dozen events.
  2. It is an RNA measurement. Reciprocal expression is an RT-PCR or RNA-seq call, so a negative is confounded with low expression and with assay design - concept-junction-rt-pcr-what-it-can-and-cannot-attribute records that the spliced-versus-reciprocal distinction is defined on RT-PCR.
  3. The direction is counter-intuitive. The patients expressing the second oncogenic product did better, which is hard to act on without a mechanism, though the recombinome's reading - that reciprocal fusion proteins act as chromatin-opening complexes required for the onset of the pre-leukaemic state and possibly dispensable afterwards - supplies one 1.
  4. Nobody has tested it as a stratifier. An absence of testing is not a negative result. Interfant-06's multivariable model did not include reciprocal status or breakpoint position, so its silence on them carries no information.

This graph has recorded across several beats that reciprocal status is stratified by nobody. This beat's contribution is to confirm that the MRD assay designers are the one group who touch it, and they touch it only when forced.

Breakpoint position: load-bearing for the assay, absent from the risk model

Breakpoint position is not optional information for a genomic MRD assay - it is the assay. The junction's location determines the restriction fragment, the primer placement and the amplicon. And the recombinome documents that position is systematically structured: an age-dependent shift with intron 11 breakpoints associated with ALL and younger patients and intron 9 breakpoints predominating in AML or older patients 2, with functional consequences, since breakpoints within intron 11 destroy the dimerisation capacity of the PHD1-3 region 1.

So the assay reads a variable with known biological structure and known functional consequences, discards it, and reports a single number. The information is already being generated in every patient who gets a breakpoint-based MRD assay: every one of those patients has their intron of origin sequenced as a by-product of assay design. No trial has linked that field to outcome prospectively, and the Interfant breakpoint-prognostic question is an open analysis in this graph (ana-interfant-breakpoint-prognostic-reanalysis, hyp-breakpoint-position-prognostic-interfant).

The cheapest version of the missing study needs no new samples: the breakpoint archive and the Interfant outcome database both exist, and the join key is the patient.

Facts and figures
questionis reciprocal status or breakpoint position used in MRD assay design or in risk stratification?
answer assay designreciprocal junction used only as a fallback when the der(11) junction cannot be characterised: 16 of 3401 patients plus 2 with 5'-KMT2A deletion and 3 of 18 intergenic-partner cases
answer risk stratificationno; neither reciprocal status nor breakpoint position appears in the Interfant-06 risk model or its multivariable analysis
breakpoint position is the assayintron position determines restriction fragment, primer placement and amplicon; every patient with a breakpoint MRD assay has their intron of origin sequenced as a by-product
missing studyjoin the KMT2A breakpoint archive to the Interfant outcome database on patient identity and test breakpoint position and reciprocal status as prognostic covariates; no new samples required
reciprocal prognostic evidence
{
  "cohort": "43 t(4;11) infants on Interfant 99/06",
  "af4_mll_expressed_n": 19,
  "af4_mll_expressed_pct": 45,
  "efs_4y_pct": {
    "expressing": 62.4,
    "non_expressing": 11.7
  },
  "efs_p": 0.001,
  "os_4y_pct": {
    "expressing": 73.7,
    "non_expressing": 25.2
  },
  "os_p": 0.016,
  "cox_hr_lacking_af4_mll": 3.42,
  "cox_ci": "1.35-8.63",
  "cox_p": 0.01,
  "cox_hr_interfant_risk_group": 1.34,
  "risk_group_p": 0.49,
  "source_pmid": "30679323"
}
reciprocal absent by architecture
{
  "kmt2a_mllt1_spliced_cases": 241,
  "of_total": 460,
  "note": "spliced fusions upstream of partner exon 1 express no reciprocal transcript",
  "source_pmid": "37019990"
}
related open work
[
  "ana-interfant-breakpoint-prognostic-reanalysis",
  "hyp-breakpoint-position-prognostic-interfant"
]
How this could bite us back
  • If reciprocal status carries a hazard ratio of 3.42 while the trial's own risk group carries 1.34 and is non-significant, then children are being allocated to transplant in first remission on a variable that was outperformed in the only analysis that compared them - and the comparison has never been repeated.
  • Building a two-target MRD assay on both derivatives would cost one extra qPCR and would detect subclonal loss of either derivative. It has never been done, so the field does not know whether single-target breakpoint MRD is missing a resistance mechanism it is structurally unable to see.
  • In the spliced-fusion architecture there is no reciprocal allele at all, so any reciprocal-based refinement is unavailable to roughly half of KMT2A::MLLT1 patients by construction. A stratifier that cannot be measured in half a partner group is not a stratifier.
  • The counter-intuitive direction - patients expressing the second oncogenic product survive better - means that acting on it would intensify therapy in the patients with *less* oncogenic material, which is the kind of rule that will not be adopted without a mechanism, however good the statistics.
Still open
  1. In patients where both the der(11) and the reciprocal junction are characterised, do the two targets give the same MRD value? If a subclone lost one derivative, they would not, and nobody has run both on the same remission marrow.
  2. Does the AF4-MLL prognostic effect replicate in a second cohort? It is the largest unexploited prognostic signal in this disease on published effect size, and it rests on 43 patients.
  3. Is the AF4-MLL effect separable from HOXA expression? The two were analysed as a composite, so the prognostic variable may be the transcriptional programme rather than the reciprocal allele.
  4. Every patient with a breakpoint-based MRD assay has their KMT2A intron sequenced. Has any trial group ever added that field to its outcome database, and if not, why not?
Every source behind this entry (5)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In 3,401 acute leukaemia patients analysed by long-distance inverse PCR, long-distance multiplex PCR and targeted NGS of full-length KMT2A, at least one patient-specific KMT2A fusion sequence was obtained per patient and used to establish patient-specific qPCR assays for MRD; for 16 patients the der(11) could not be identified but a reciprocal fusion allele was diagnosed, two patients had a 5'-KMT2A deletion with reciprocal fusion genes, and of 18 patients with no partner gene fused to 5'-KMT2A a reciprocal fusion gene was identified in three; of 426 patients with complex rearrangements, 40 reciprocal KMT2A fusions were in-frame and 386 were non-functional or out-of-frame, the majority of which may still be transcribed to encode the 5'-truncated KMT2A* protein from a gene-internal promoter upstream of KMT2A exon 12; about 50% of KMT2A::MLLT1 cases (241 of 460) break upstream of MLLT1 exon 1 and express no reciprocal transcript; and breakpoints within KMT2A intron 11 destroy the dimerisation capacity of the PHD1-3 region.

    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. A referral-based diagnostic series, so proportions are not population estimates. The counts of reciprocal-only cases are small absolute numbers in a large series and the paper does not state whether those patients' MRD assays were actually built on the reciprocal junction - only that the reciprocal allele was what was characterised. The KMT2A* inference for out-of-frame reciprocal fusions is structural, from promoter position, not measured in those patients.

    Evidence that argues against this (1)
    • The paper also states that RNA-Seq provides insight into alternative splice events that genomic analysis cannot supply, explicitly for out-of-frame reciprocal cases, so a DNA-only characterisation of reciprocal status is incomplete on the authors' own account. Meyer C et al. 2023
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Analysis of 2,345 acute leukaemia patients showed an age-dependent KMT2A breakpoint shift with breakpoints within intron 11 associated with acute lymphoblastic leukaemia and younger patients while breakpoints in intron 9 predominate in acute myeloid leukaemia or older patients, and the authors state that establishing patient-specific chromosomal fusion sites allows the design of specific PCR primers for minimal residual disease analyses for all 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 network and largely the same pipeline as E1, so these are not independent series. The age-lineage breakpoint association is cross-sectional and confounded by the strong association between age, lineage and fusion partner.

    Evidence that argues against this (1)
    • The 2023 update from the same network found that the intron 11 breakpoint shift in adults with MLLT3 fusions was clearly linked to therapy-induced leukaemia, and was not seen for ELL, EPS15 or SEPTIN6 fusions - so the age-dependent shift is partner- and aetiology-specific rather than a uniform property of the locus. Meyer C et al. 2023
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Interfant-06 defined its three risk groups on KMT2A rearrangement status, age at diagnosis relative to 6 months, white blood cell count relative to 300 x10^9/L and prednisone response, and reported KMT2A rearrangement as the strongest prognostic factor for event-free survival followed by age, WBC count and prednisone response.

    PMID 31283407DOI
    What this evidence cannot show

    Abstract only; the full text is not open access and no PMCID is returned by the Europe PMC core record. The abstract states the ordering of prognostic factors without effect sizes, and it does not state which candidate variables were offered to the multivariable model - so that breakpoint position and reciprocal status were absent is an inference from their non-appearance, i.e. an absence of testing rather than a negative result.

    Evidence that argues against this (1)
    • An absence from a published multivariable model does not establish that the variable was tested and rejected. The Interfant-06 dataset may never have carried a breakpoint-position field, in which case the model could not have included it. Meyer C et al. 2023
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Among 124 de novo infant B-ALL cases uniformly diagnosed and treated on the Interfant 99/06 protocol, the reciprocal fusion AF4-MLL was expressed in only 45% (19 of 43) of t(4;11)-positive patients, its expression correlated with HOXA cluster expression (P = 0.002), AF4-MLL/HOXA-expressing patients had 4-year event-free survival 62.4% versus 11.7% (P = 0.001) and overall survival 73.7% versus 25.2% (P = 0.016), and in a Cox model adjusting for medium- versus high-risk stratification according to the Interfant-06 protocol the hazard ratio for patients lacking AF4-MLL expression was 3.42 (95% CI 1.35-8.63, P = 0.01) while risk group was not significant (HR 1.34, 95% CI 0.59-3.03, P = 0.49).

    PMID 30679323DOI
    What this evidence cannot show

    Read in full text. 43 t(4;11) patients with wide standard errors on both survival curves (SE 10-11 percentage points). Reciprocal expression is an RNA-level call, so a negative conflates absence of the allele, absence of transcription and assay sensitivity. AF4-MLL expression and HOXA expression are treated as a single composite in the survival analysis, so the prognostic signal cannot be attributed to the reciprocal fusion alone. Single study, not replicated.

    Evidence that argues against this (1)
    • Reciprocal status is not an independent variable: the recombinome shows that the cases expressing no reciprocal transcript are precisely those with a spliced-fusion breakpoint upstream of the partner's exon 1 - 241 of 460 for MLLT1, 1 of 1262 for AFF1, 2 of 612 for MLLT3. Reciprocal status is therefore partly a relabelling of breakpoint architecture, and in t(4;11) the spliced architecture is so rare that AF4-MLL non-expression must have a different cause, most likely transcriptional. Meyer C et al. 2023
  5. E5strong 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 used as the main MRD-PCR target with immunoglobulin and T-cell receptor gene rearrangements as additional targets, the breakpoint fusion sequence was determined at one of two reference centres, and patient-specific primers were designed on the basis of the junctional region of the identified genomic rearrangement.

    PMID 33405950DOI
    What this evidence cannot show

    Read in full text. The description says 'the KMT2A genomic breakpoint fusion sequence' in the singular and does not state whether reciprocal junctions were ever used, so the absence of a two-target strategy is an inference from the wording rather than an explicit statement that none was used.

    Evidence that argues against this (1)
    • The same paper reports that where both KMT2A and IG/TR targets existed the higher of the two results was used for analysis, so the protocol was willing to combine targets in principle - which makes the absence of a reciprocal target a choice rather than a methodological impossibility. Stutterheim J et al. 2021
Connected to