Oncofusion

start / graph / concept-proximal-vdj-bias-and-ig-mrd-in-kmt2a-aff1

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

Proximally biased V(D)J and IGHV6-1 in KMT2A::AFF1: the Ig MRD target is a moving target, and the clinic had already moved

also called IGHV6-1 bias · proximal IGHV usage · DNJH stems · ongoing DJH recombination · clonal evolution of IGH in KMT2A-r

In this leukaemia the antibody gene is rearranged in an unusually restricted and unfinished way: the cells overwhelmingly use the gene segment physically closest to the joining region, and many of them are still actively rearranging. Because the standard leftover-disease test reads that rearrangement as the cancer's fingerprint, the fingerprint can keep changing while treatment is going on.

The same thing, for a scientist

Whole-transcriptome sequencing of 379 BCP-ALL cases, including 48 adult, 36 paediatric and 37 infant t(4;11)/KMT2A::AFF1 cases, found strongly favoured usage of **proximal** IGHV genes, with the most proximal gene IGHV6-1 showing significantly higher expression than distal ones in KMT2A-rearranged samples of all age classes; up to 50% of KMT2A-rearranged samples harboured B-cell receptors involving IGHV6-1 [E1]. The same samples showed a median of **three different DNJH stems per sample** - matched only by t(1;19)/TCF3::PBX1, and contrasting with the DUX4 subtype where multiple DNJH stems were not detected - and shorter nucleotide-level CDR3 regions than most other subtypes [E1]. The authors read the pattern as more abundant de novo DJH recombination and a more primitive differentiation stage, consistent with IGH locus decontraction being controlled by PAX5 and WAPL, and state that these findings have implications for clinical management because clonal evolution interferes with reliable detection of MRD based on IG rearrangements; their suggested alternative is detection of fusion products and co-mutations [E1]. An independent adult cohort of 280 BCP-ALL patients found IGH clonal evolution at diagnosis in one third, with high prevalence and degree of evolution driven by ongoing DH/VH-DJH recombination **associated with KMT2A rearrangement** [E2]. **The clinical correction matters as much as the finding**: Interfant-06 had already made the genomic KMT2A breakpoint the preferred MRD target, on the stated ground that IG/TR rearrangements here are occasionally absent, oligoclonal or less stable [E3]. This work supplies a mechanism for a switch made three years earlier, and extends it to adults and older children, who are still monitored on Ig/TCR.

Go deeper
What was measured, precisely

Read in full text. MiXCR was run on RNA-seq reads and IGHV expression quantified as reads mapping to each IGHV gene, normalised per sample; genes were ordered by genomic distance to the DJH cluster on chromosome 14, IGHV6-1 being the most proximal and IGHV3-76 the most distal shown. For most samples IGHV expression was independent of genomic position; in KMT2A-rearranged samples IGHV6-1 was significantly higher than distal genes, and this held in all three age classes 1.

Junction analysis used IMGT/V-QUEST, which maps the nucleotides encoding the CDR3 anchor positions C104-W118 to VH-N-D-N-JH components. KMT2A-rearranged samples of all age classes had a median of three DNJH stems per sample; only TCF3::PBX1 was comparable; DUX4 showed no multiple stems. CDR3 regions in KMT2A-rearranged samples were shorter at the nucleotide level than in most other subtypes 1.

Three things the study is not: it is not longitudinal, no remission sample was tested; it is not DNA-based, so the quantity is transcript abundance rather than clone size; and no MRD assay was run or observed to fail. The MRD claim in the paper is an inference the authors draw, stated as an implication.

Why a proximal bias and ongoing recombination should hurt an Ig MRD assay

Three separable mechanisms, each with a different consequence:

  1. Ongoing DJH recombination means the clone is still rearranging. A patient-specific ASO primer is designed against one junctional sequence. If the clone continues to recombine during therapy, descendants of the MRD-relevant cell may not carry the sequence the primer was designed for. This is the mechanism the Interfant description of less stable targets refers to 3, and the mechanism the adult HTS cohort attributes specifically to KMT2A-rearranged disease 2.
  2. Multiple DNJH stems mean oligoclonality is the rule, not the exception. A median of three per sample 1 is consistent with the older finding that Ig/TCR rearrangements in infant ALL are more oligoclonal than in childhood precursor-B ALL 4. An assay built on one of three coexisting stems will track one third of the disease, and therapy chooses which third survives.
  3. Proximal bias compresses the design space and raises the background. If half of KMT2A-rearranged samples use IGHV6-1 1, patient-specific assays in this subtype concentrate on one V segment. That is a problem for specificity rather than for stability: the discriminating sequence is then almost entirely the N-D-N junction, and CDR3 regions in this subtype are shorter 1, so there is less junctional sequence to discriminate with. Shorter CDR3 plus shared V usage is the worst combination for allele-specific primer design. This last inference is ours, not the paper's, and is stated as a prediction rather than a finding.
The correction to the framing

It would be easy to present this as a new threat to clinical practice. It is better described as a mechanism arriving after the practice had already changed, and as a warning aimed at the populations where the practice has not changed.

  • Infants: Interfant-06 already used the genomic KMT2A breakpoint as the main MRD target, with Ig/TR as additional 3. The concern is largely pre-empted.
  • Adults and older children with KMT2A-rearranged BCP-ALL: these are 48 and 36 of the cases in this study 1, and they are not monitored on an Interfant protocol. Adult ALL MRD is Ig/TCR- or flow-based as standard 5, and the adult HTS cohort found the clonal-evolution association with KMT2A rearrangement in adults 2. This is where the finding is actionable and where nothing has changed.

The authors' own suggested alternative - detection of fusion products and co-mutations 1 - is exactly the genomic-breakpoint and fusion-transcript route this beat documents. The implication for the graph is therefore not that Ig MRD is newly suspect, but that the age group in which Ig MRD is still standard is the age group in which the mechanism was most strongly measured.

Facts and figures
methodMiXCR on RNA-seq reads for IGHV expression; IMGT/V-QUEST junction analysis (C104-W118 anchors)
proposed mechanismIGH locus decontraction for a randomised repertoire depends on PAX5 and its target WAPL; proximal bias plus de novo DJH recombination suggest a more primitive differentiation stage
authors suggested alternativedetection of fusion products and co-mutations
cohort
{
  "total_bcp_all": 379,
  "kmt2a_aff1_adult": 48,
  "kmt2a_aff1_paediatric": 36,
  "kmt2a_aff1_infant": 37,
  "source_pmid": "38650597"
}
findings
{
  "most_proximal_ighv": "IGHV6-1",
  "most_distal_shown": "IGHV3-76",
  "samples_with_ighv6_1_bcr_pct_upper": 50,
  "median_dnjh_stems_per_sample": 3,
  "only_comparable_subtype": "t(1;19)/TCF3::PBX1",
  "subtype_with_no_multiple_stems": "DUX4",
  "cdr3_length": "shorter at nucleotide level than most other BCP-ALL subtypes"
}
what was not done
[
  "no remission samples",
  "no MRD assay run or observed to fail",
  "no DNA-based orthogonal validation of stem counts",
  "no longitudinal tracking of a clone through therapy"
]
clinical status
{
  "infants": "Interfant-06 already uses the genomic KMT2A breakpoint as main target",
  "adults_and_older_children": "Ig/TCR and flow remain standard; no change"
}
How this could bite us back
  • MRD drives intensification and transplant referral in adult ALL. If Ig/TCR MRD systematically under-reads in KMT2A::AFF1 adults - the population this study measured most heavily and the one whose practice has not changed - then under-treatment is concentrated in the subtype with the worst prognosis.
  • The obvious fix, switching those patients to fusion-based MRD, requires a breakpoint characterisation service that adult haematology outside trial networks largely does not have. Recommending the better target without the infrastructure converts a measurement problem into an access problem.
  • A clone that keeps rearranging can also lose the epitope a CD19-directed therapy needs, and both are consequences of an immature, still-recombining B-lineage state. The assay problem and the antigen-escape problem may share a cause, in which case the patients whose MRD target is least stable are the ones for whom immunotherapy failure is likeliest.
  • The proximal bias is measured on RNA-seq read counts, so a therapy that changes IGH transcription without changing clone composition would change the measured bias. Anyone using this as a monitoring biomarker rather than a diagnostic feature would be reading transcription as clonality.
Still open
  1. Does an ASO-PCR target designed against an IGHV6-1-using clone in KMT2A::AFF1 actually fail more often during therapy than one against a distal V segment? This is answerable from any EuroMRD laboratory's archive and has not been asked.
  2. Does shorter CDR3 plus concentrated IGHV6-1 usage degrade ASO primer specificity in this subtype? The prediction follows from the data but the paper did not test it and no MRD laboratory has reported design failure rates by subtype.
  3. Is the proximal bias a property of the cell of origin or an effect of the fusion on IGH locus contraction? The PAX5/WAPL mechanism the authors cite is testable: measure IGH locus contraction and WAPL expression in KMT2A::AFF1 blasts against subtype-matched controls.
  4. Adults with KMT2A::AFF1 are 48 of the cases in this study and are monitored on Ig/TCR. Has anyone looked at whether their Ig/TCR MRD discordance with fusion-based MRD resembles the infant pattern?
Every source behind this entry (5)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In 48 adult, 36 paediatric and 37 infant BCP-ALL cases carrying t(4;11)(q21;q23)/KMT2A::AFF1, whole-transcriptome sequencing with MiXCR showed strongly favoured usage of proximal IGHV genes with the most proximal gene IGHV6-1 significantly higher-expressed than distal genes in all age classes, up to 50% of KMT2A-rearranged samples harbouring B-cell receptors involving IGHV6-1, a median of three different DNJH stems per sample, shorter nucleotide-level CDR3 regions than most other BCP-ALL subtypes, and the authors state that clonal evolution interferes with reliable detection of MRD based on IG rearrangements and that an obvious alternative would be detection of fusion products and co-mutations.

    PMID 38650597DOI
    What this evidence cannot show

    Read in full text via Europe PMC fullTextXML. Cross-sectional diagnostic samples only; no remission sample was analysed and no MRD assay was run, so the MRD implication is inference. The measure is RNA-seq read counts per IGHV gene normalised per sample, which conflates clone frequency with per-cell transcript abundance. Most samples come from one commercial diagnostic laboratory (Munich Leukemia Laboratory) with part-ownership declared by three senior authors, plus a validation cohort contributed from St Jude. The DNJH stem count depends on the MiXCR and IMGT/V-QUEST pipeline and no orthogonal DNA-based validation is presented.

    Evidence that argues against this (2)
    • Interfant-06 had already adopted the genomic KMT2A breakpoint as the preferred MRD target three years earlier, stating that IG/TR rearrangements in this disease are occasionally absent, oligoclonal or less stable; the practice implication had therefore already been acted on in the infant population where the disease is commonest. Stutterheim J et al. 2021
    • Deep BCR repertoire sequencing of t(4;11) infant samples from a different group found no significantly expanded VDJ-rearranged clones at all, which is a different picture from a biased but present repertoire and suggests the two methods are measuring different quantities. Agraz-Doblas A et al. 2019
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In 280 adults with newly diagnosed BCP-ALL studied by high-throughput sequencing with clonally related IGH sequences identified by shared DNJ-stem, IGH clonal evolution at diagnosis was present in one third of patients, and in 167 patients with molecular subtype allocation high prevalence and high degree of clonal evolution driven by ongoing DH/VH-DJH recombination was associated with the presence of KMT2A gene rearrangements, while VH replacements occurred more frequently in Ph-like and DUX4 BCP-ALL.

    PMID 37143651DOI
    What this evidence cannot show

    Adults only; the KMT2A-rearranged subgroup within the 167 subtyped patients is not large and its size is not given in the abstract. Clonal evolution is inferred from sequence relatedness at diagnosis, not observed over time. The DNJ-stem concept is introduced by this group and used by the KMT2A::AFF1 study as its comparator, so the two analyses share a framework and are not fully independent.

    Evidence that argues against this (1)
    • The same paper identifies VH replacement as a second, distinct clonal-evolution mechanism that is not enriched in KMT2A-rearranged disease, so KMT2A-r evolution is one of at least two processes and the MRD consequences of each differ. Darzentas F et al. 2023
  3. E3established 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, on the stated ground that earlier studies obtained more reliable MRD data with the KMT2A marker than with the occasionally absent, oligoclonal or less stable IG/TR rearrangements.

    PMID 33405950DOI
    What this evidence cannot show

    Read in full text. The justification is the authors' summary of earlier work rather than a comparison performed in that paper; the head-to-head data they do report show 94% concordance and no significant overall discrepancy between the two targets.

    Evidence that argues against this (1)
    • The trial's own agreement analysis found no significant discrepancy between the two target types (P = .60 at end of induction), so the stated reliability advantage is not visible in the trial's primary comparison and emerges only at the decision cut-off. Stutterheim J et al. 2021
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In 133 infants with ALL, Ig/TCR rearrangements were less frequent and more oligoclonal than in childhood precursor-B ALL, MLL-AF4 and MLL-ENL-positive infants demonstrated immature rearrangements, and the immature pattern correlated with young age at diagnosis, CD10-negativity and predominantly with the presence and type of MLL translocation.

    PMID 17268512DOI
    What this evidence cannot show

    Mid-2000s detection methods, cross-sectional, and oligoclonality is a categorical call rather than a count of stems, so it cannot be compared numerically with the DNJH stem medians in E1.

    Evidence that argues against this (1)
    • The same study found IGH rearrangements in 91% of infants, so the majority do have a candidate target; oligoclonality degrades target quality rather than abolishing it. Jansen MW et al. 2007
  5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Flow cytometry and PCR analysis of rearranged immunoglobulin and T-cell receptor genes are the two established MRD techniques in ALL, used in frontline treatment of virtually all childhood and many adult ALL patients for risk-group assignment.

    PMID 25999452DOI
    What this evidence cannot show

    A review, and one written by the developers of the standard methods, so it describes the standard of care rather than independently auditing it. It does not break the figure down by subtype or by age within adults.

    Evidence that argues against this (1)
    • A 2021 consensus exercise in one Canadian province found great variability in where, when and by which technique adult B-ALL MRD is performed and interpreted, so a description of the standard is not a description of practice. Chen D et al. 2021
Connected to