Oncofusion

start / graph / concept-cytogenetics-karyotype-kmt2a

Explainerstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q7-clinical

Step 3: the G-banded karyotype and t(4;11)(q21;q23)

also called karyotype · conventional cytogenetics · G-banding · chromosome analysis · t(4;11) · 11q23 translocation

Cytogeneticists grow the leukaemia cells for a day or two until they divide, freeze them mid-division, stain the chromosomes into stripes and look at them under a microscope. In the classic case they see that a piece of chromosome 11 and a piece of chromosome 4 have swapped places - that is t(4;11), the picture that named this disease. It is the only test that shows the whole genome at once, and it is also the test that fails most often in babies.

The same thing, for a scientist

Conventional G-banded karyotyping of the diagnostic marrow remains the only single assay that surveys the entire genome for structural and numerical change, and it is where t(4;11)(q21;q23) - the cytogenetic signature of KMT2A::AFF1 - is named. It requires 24-48 hours of unstimulated culture to accumulate metaphases, and it has systematic blind spots that matter specifically in infant leukaemia: culture failure and poor chromosome morphology, overgrowth by normal dividing cells while blasts remain non-dividing, cryptic insertions and complex multi-way rearrangements that place KMT2A sequence at a cytogenetically invisible position, and KMT2A partial tandem duplications that are intragenic and therefore below banding resolution [E3][E5]. In a 51-infant series, 13 (25%) had contradictory cytogenetic and Southern-blot results, and FISH resolved these as cryptic insertion or translocation in five and as normal dividing cells with non-dividing leukaemic cells in five [E3]. The long tail of the recombinome makes the point structurally: across 3401 patients, 107 in-frame KMT2A fusions have been catalogued, along with 18 patients with no partner gene fused to 5'-KMT2A at all and two with a 5'-KMT2A deletion [E1]. Karyotype therefore establishes the named translocation and the wider clonal context, and must be paired with FISH and molecular testing rather than trusted alone [E4].

Go deeper
What the karyotype is, and what t(4;11) looks like on it1 deeper

A karyotype is a photograph of one cell's chromosomes, caught in metaphase, stained so that each chromosome carries a reproducible pattern of light and dark bands, and then arranged in order by size. It is the oldest assay in this pathway and still the only one that answers the question what else is going on in this genome without being told in advance what to look for.

In KMT2A::AFF1 leukaemia the finding is a reciprocal translocation written in ISCN nomenclature as 46,XX,t(4;11)(q21;q23) or 46,XY,t(4;11)(q21;q23). Read it literally: two chromosomes, 4 and 11, have exchanged material; the breakpoints sit at band q21 on chromosome 4 and band q23 on chromosome 11. On the banded spread the derivative chromosome 11 is visibly short in its long arm and the derivative chromosome 4 visibly long, and the band pattern distal to each breakpoint is the wrong one for that chromosome. The der(11) is the product that matters biologically - it carries 5'-KMT2A joined to 3'-AFF1 - although the karyotype itself cannot tell you which derivative encodes the expressed fusion.

t(4;11)/KMT2A::AFF1 is one of the abnormalities with the most significant impact on the treatment and management of B-cell precursor ALL and is used for high-risk stratification, which is why naming it correctly on the karyotype has immediate consequences for the protocol arm 4.

Why it takes 24 to 48 hours

A chromosome only becomes visible when a cell condenses it to divide. Everything in the karyotype workflow follows from that constraint:

  1. Marrow is cultured, conventionally unstimulated and harvested at both 24 and 48 hours so that two independent windows of division are sampled.
  2. Colcemid or a similar spindle poison is added to arrest cells in metaphase and accumulate them.
  3. Hypotonic treatment swells the cells, fixative is applied, and the suspension is dropped onto slides so that the chromosomes spread without overlapping.
  4. Trypsin-Giemsa banding produces the stripe pattern.
  5. A cytogeneticist analyses, conventionally, 20 metaphases - imaging, karyotyping and interpreting each one.

Two working days is therefore the floor, not the typical case. Slow-growing or low-viability samples extend it, and every step after culture is manual. This is the reason the karyotype cannot be the test that decides the first day of treatment, and the reason flow cytometry and FISH sit in front of it in the diagnostic sequence.

How the karyotype fails, and how often1 deeper

The failure modes are not exotic. They were quantified directly in infant leukaemia. Of 51 infants with acute leukaemia analysed by both cytogenetics and Southern blot, 13 (25%) had contradictory findings on 11q23/KMT2A rearrangement status 3:

DiscordancenWhat it means
Rearranged KMT2A, normal karyotype7the break is real and the banded spread shows nothing
Rearranged KMT2A, abnormal karyotype with no 11q23 translocation4something structural is visible, but not at the right place
Germline KMT2A, 11q23 translocation present2the visible 11q23 break is not in KMT2A

FISH with a KMT2A probe resolved these cases into two mechanisms, each accounting for five of the thirteen: normal dividing cells coexisting with non-dividing leukaemic cells in the same marrow, and cryptic insertion or translocation 3. Subsequent FISH and RT-PCR named MLL-AF10, MLL-AF4 or MLL-AF1q fusions produced by the cryptic rearrangements in four of those five. In one remaining patient the visible 11q23 breakpoint lay distal to the KMT2A locus; in two the discrepancy was never resolved.

Unpack the first mechanism, because it is the one people forget. A karyotype does not sample the marrow; it samples the cells in the marrow that divided in culture. If the blast clone is quiescent and the residual normal progenitors are not, twenty beautifully banded metaphases can all be normal cells from a marrow that is 95% leukaemia. The result is not a false negative in the assay's own terms - every metaphase really was normal - and that is exactly what makes it dangerous.

What is structurally invisible to banding

Beyond sampling, some rearrangement classes cannot be seen at banding resolution in principle:

  • Cryptic insertions. A segment of KMT2A, or of the partner, is inserted into the other chromosome rather than exchanged reciprocally. The insert is too small to perturb the band pattern, so both chromosomes look normal 3.
  • Complex and multi-way rearrangements. The recombinome catalogue documents reciprocal, complex and therapy-induced translocations as distinct classes, and an expanding network of reciprocal KMT2A fusions arising from complex rearrangements 2. A three-way exchange can leave each individual chromosome looking only mildly odd.
  • Rearrangements with no partner gene. Of 3401 patients characterised at the genomic breakpoint, 18 had no partner gene fused to 5'-KMT2A and two had a 5'-KMT2A deletion; one ETV6::RUNX1 patient had a KMT2A insertion at the breakpoint 1. There is nothing for a translocation-shaped search to find in these cases.
  • Partial tandem duplications (KMT2A-PTD). These are small intragenic insertions, explicitly described as particularly challenging to detect using standard molecular and cytogenetic approaches 5. In 932 AML cases, 41 harboured a PTD, and establishing them required cross-checking targeted NGS against MLPA and optical genome mapping; unbalanced chromosomal rearrangements overlapping KMT2A actively interfered with copy-number-based PTD detection, producing false-positive calls 5. No banded karyotype resolves an event inside a single gene.

The scale of the long tail is the deeper point. Seven partner genes plus PTDs account for more than 90% of all KMT2A recombinations, but 37 rearrangements recur and 63 have been seen only once 1. A diagnostic strategy tuned to the common, cytogenetically obvious translocations will be right most of the time and systematically wrong about the rare cases.

What the karyotype still uniquely provides

None of the above argues for dropping it. The karyotype is the only assay in the panel that is not hypothesis-directed. FISH looks where the probe is. RT-PCR finds what its primers encode. The karyotype looks everywhere at once, at coarse resolution, and returns three things nothing else in the routine pathway gives:

  1. The partner chromosome, named. A break-apart FISH result says KMT2A is broken; the karyotype says it went to 4q21, which is most of the way to naming the fusion.
  2. Additional clonal abnormalities and ploidy. Near-haploidy and low hypodiploidy are themselves high-risk stratifiers in B-cell precursor ALL, and high hyperdiploidy a good-risk one, and these are numerical findings a targeted probe set will not report 4.
  3. Clonal architecture. Multiple related abnormal clones, or an abnormal clone in a fraction of metaphases, describe a disease's structure in a way a binary positive/negative result cannot.

The international guidance is correspondingly pragmatic: the prognostically decisive abnormalities in B-cell precursor ALL, including t(4;11)(q21;q23)/MLL-AFF1, can be routinely screened by RT-PCR, while cytogenetics and particularly FISH provide reliable alternative detection depending on the technical approach each protocol prefers 4. That is not an endorsement of any one assay - it is an instruction to run more than one.

Facts and figures
assayG-banded karyotype (conventional cytogenetic analysis)
specimenbone marrow aspirate, sterile, heparinised; peripheral blood only if blasts are circulating in quantity
culture typeunstimulated, colcemid-arrested
metaphases conventionally scored20
canonical iscn46,XX,t(4;11)(q21;q23) or 46,XY,t(4;11)(q21;q23)
derivative carrying fusionder(11), encoding 5'-KMT2A::3'-AFF1
risk rolet(4;11)/KMT2A::AFF1 is a high-risk stratifier in childhood B-cell precursor ALL
culture hours
[
  24,
  48
]
turnaround working days
[
  2,
  7
]
blind spots
[
  {
    "mechanism": "non-dividing blasts, dividing normal cells",
    "effect": "all metaphases normal despite a heavily infiltrated marrow",
    "observed_n": "5 of 13 discordant infants (Watanabe 2003)"
  },
  {
    "mechanism": "cryptic insertion or translocation",
    "effect": "band pattern unchanged; KMT2A sequence relocated below banding resolution",
    "observed_n": "5 of 13 discordant infants (Watanabe 2003)"
  },
  {
    "mechanism": "complex or multi-way rearrangement",
    "effect": "no single chromosome looks clearly abnormal; reciprocal fusion network documented in the recombinome"
  },
  {
    "mechanism": "no partner gene fused to 5'-KMT2A, or 5'-KMT2A deletion",
    "effect": "nothing translocation-shaped to find; 18 and 2 patients respectively of 3401 (Meyer 2023)"
  },
  {
    "mechanism": "KMT2A partial tandem duplication",
    "effect": "intragenic; invisible to banding and to a break-apart probe"
  },
  {
    "mechanism": "culture failure or poor chromosome morphology",
    "effect": "no analysable result; more frequent in infants where aspirate volume is small and marrow is packed"
  }
]
unique value
[
  "names the partner chromosome and band, not just that KMT2A is broken",
  "reports ploidy, including near-haploidy/low hypodiploidy and high hyperdiploidy",
  "reports additional clonal abnormalities and clonal architecture",
  "is not hypothesis-directed: no probe or primer has to be chosen in advance"
]
discordance with molecular testing
{
  "cohort": "51 infants with acute leukaemia",
  "discordant_n": 13,
  "discordant_pct": 25,
  "source": "Watanabe 2003, PMID 12750700"
}
How this could bite us back
  • A normal karyotype reported without qualification invites the clinical team to read 'no KMT2A rearrangement'. In the best-quantified infant series a quarter of cases had cytogenetic and molecular results that disagreed, so a normal karyotype alone would mis-stratify a high-risk infant into a standard-risk arm.
  • The karyotype samples only cells that divided in culture. If the blast clone is quiescent, twenty perfectly normal metaphases can come from a marrow that is almost entirely leukaemia - and every individual metaphase report will be accurate.
  • Culture failure does not produce a wrong answer, it produces no answer; but the risk group still has to be assigned on the day, so the practical consequence is a decision made on incomplete data.
  • If the karyotype is treated as the definitive genetic test, a KMT2A partial tandem duplication or a cryptic insertion is missed entirely, and the child is entered on a protocol arm - and potentially excluded from a menin-inhibitor trial - on the basis of an assay that could not have seen the lesion.
  • The 90/10 structure of the recombinome is itself a trap: because seven partners plus PTDs cover more than 90% of cases, a pathway optimised for the common translocations performs superbly on audit while being systematically blind to the 63 partners seen only once.
  • Naming the translocation correctly matters in the other direction too: an 11q23 break visible on banding is not necessarily in KMT2A - two infants in the Watanabe series had visible 11q23 translocations with germline KMT2A - so a cytogenetics-led diagnosis can also over-call the lesion.
Still open
  1. What fraction of KMT2A rearrangements in a modern consecutive infant cohort is cytogenetically invisible when FISH plus RNA sequencing is the reference standard? The best available figure - 25% discordance - comes from a 51-infant series from 2003 with Southern blot as the comparator, and conflates true cryptic rearrangement with culture sampling artefact.
  2. What is the current karyotype failure rate specifically in infants under one year, and does it differ from older children once aspirate volume and marrow cellularity are accounted for?
  3. Does optical genome mapping, which resolved KMT2A rearrangements that MLPA could not in adult AML, replace or merely supplement the karyotype's genome-wide role in infant ALL? Nobody appears to have run that comparison in an infant cohort.
  4. Of the 63 KMT2A rearrangements so far seen only once, how many were found only because LDI-PCR was applied after a normal or uninformative karyotype - and how many comparable cases were never characterised because that step was never taken?
Every source behind this entry (5)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Across 3401 acute leukaemia patients analysed between 2003 and 2022, genomic breakpoint characterisation identified 107 in-frame KMT2A gene fusions in total, plus 16 out-of-frame fusions, 18 patients with no partner gene fused to 5'-KMT2A, two patients with a 5'-KMT2A deletion and one KMT2A insertion at an ETV6::RUNX1 breakpoint; the seven most frequent partner genes plus partial tandem duplications account for more than 90% of all KMT2A recombinations, 37 rearrangements recur and 63 have been identified only once.

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

    A referral-based diagnostic series accumulated over 20 years, not a population sample: samples arrive at the Frankfurt Diagnostic Center of Acute Leukemia precisely because a KMT2A rearrangement is suspected, so partner frequencies reflect referral patterns and the LDI-PCR method's reach. Rearrangements whose breakpoints lie outside the KMT2A breakpoint cluster region are under-represented by construction. The catalogue counts rearrangements, not the fraction of karyotypes that miss them.

    Evidence that argues against this (1)
    • A count of characterised fusions is a count of what the assay could see. The same series records 18 patients with no partner gene fused to 5'-KMT2A and 16 out-of-frame fusions, which are the visible edge of that limit; spliced fusions, where the break lands upstream of an intact partner, are recovered only if RNA is examined. Karyotype in particular misses cryptic rearrangements, so '107 in-frame fusions' is a floor on partner diversity and not an enumeration of it. Leukemia 2023
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    In 2345 acute leukaemia patients, 135 different MLL/KMT2A rearrangements had been identified with 94 partner genes characterised at the molecular level, 35 of which recur, while nine specific gene fusions accounted for more than 90% of illegitimate recombinations; breakpoints showed an age-dependent shift, with KMT2A intron 11 breakpoints associated with ALL and younger patients and intron 9 breakpoints predominating in AML and older patients, and the study documented an expanding network of reciprocal MLL fusions arising from complex rearrangements.

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

    Same referral-cohort limitation as the 2023 update, and the two papers share patients, so they are not independent confirmations of each other. The intron 9 versus intron 11 age shift is an association across a heterogeneous cohort and the paper does not establish a mechanism. Breakpoint assignment depends on LDI-PCR succeeding, which it does not always do.

    Evidence that argues against this (1)
    • The partner counts are ascertainment-limited in a way that biases the concentration claim upward. Nine fusions accounting for more than 90% of recombinations is measured in a cohort characterised largely by karyotype and targeted PCR, which detect the common partners by design and find rare ones only when someone looks; the rare tail is therefore undercounted. The 94 partners known at that point have since grown. Blood Cells, Molecules and Diseases 2010
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Of 51 infants with acute leukaemia, 13 (25%) had contradictory cytogenetic and Southern-blot findings for 11q23/MLL rearrangement - seven with rearranged MLL and a normal karyotype, four with rearranged MLL and an abnormal karyotype lacking an 11q23 translocation, and two with germline MLL despite an 11q23 translocation; FISH with an MLL probe attributed five of these to normal dividing cells coexisting with non-dividing leukaemic cells and five to cryptic insertion or translocation, and subsequent FISH/RT-PCR identified MLL-AF10, MLL-AF4 or MLL-AF1q fusions in four of the latter five.

    Cryptic insertion and translocation or nondividing leukemic cells disclosed by FISH analysis in infant acute leukemia with discrepant molecular and cytogenetic findings · Watanabe N, Kobayashi H, Ichiji O, Yoshida MA, Kikuta A, Komada Y, Sekine I, Ishida Y, Horiukoshi Y, Tsunematsu Y, Yano M, Nakadate H, Kaneko Y · Leukemia 2003
    PMID 12750700DOI
    What this evidence cannot show

    51 infants from one Japanese collaboration in the pre-genomic era, and the reference standard was Southern blot, which itself has a false-negative rate for breaks outside the probed region. The 25% discordance rate is a discordance rate between two imperfect assays, NOT a measured rate of cytogenetically cryptic KMT2A rearrangement - the same series shows that some discordances were sampling artefacts rather than cryptic rearrangements. Whether 25% still holds with modern culture technique and higher-band-resolution analysis is unknown.

    Evidence that argues against this (1)
    • FISH is not a guaranteed rescue for a normal karyotype. In 38 AML cases with no 11q23 translocation on conventional cytogenetics, interphase FISH with an MLL probe found no MLL rearrangement in any case, and the authors concluded that interphase FISH 'seemed less useful for the detection of MLL rearrangement' compared with its yield for AML1/ETO, PML/RARa and CBFB/MYH11. Bai SX et al. 2011
  4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.review

    t(4;11)(q21;q23)/MLL-AFF1 is one of the abnormalities with the most significant impact on the treatment and management of B-cell precursor ALL and is used for high-risk stratification alongside t(9;22)/BCR-ABL1 and near-haploidy/low hypodiploidy; these can be routinely screened by RT-PCR as a basic yet informative screen, while cytogenetics and particularly FISH provide reliable alternative detection methods depending on the technical approach of each protocol.

    PMID 20701601DOI
    What this evidence cannot show

    Consensus recommendations from 2010, written before routine RNA sequencing and optical genome mapping, and framed for childhood B-cell precursor ALL generally rather than for infants specifically. The document deliberately allows each protocol to choose its technical approach, so it does not settle which assay should be primary; it also does not quantify the miss rate of any individual method.

    Evidence that argues against this (1)
    • RT-PCR as a basic screen has a specific blind spot for this rearrangement: it detects the transcripts its primers were designed for. A panel covering t(4;11) will miss a KMT2A fusion to any of the dozens of rarer partners, and the patient is then reported as germline KMT2A rather than as untested for the partner. Leukemia 2001
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    KMT2A partial tandem duplications are small intragenic insertions that are particularly challenging to detect using standard molecular and cytogenetic approaches; among 932 AML cases 41 harboured a KMT2A-PTD, confirmation required cross-checking targeted NGS against MLPA and optical genome mapping, and unbalanced chromosomal rearrangements overlapping KMT2A interfered with copy-number-based PTD detection and produced false-positive calls.

    Genomic Characterization of Partial Tandem Duplication Involving the KMT2A Gene in Adult Acute Myeloid Leukemia · Seto A, Downs G, King O, Salehi-Rad S, Baptista A, Chin K, Grenier S, Nwachukwu B, Tierens A, Minden MD, Smith AC, Capo-Chichi JM · Cancers 2024
    PMID 38730645DOI
    What this evidence cannot show

    An ADULT AML cohort from a single Canadian centre, so the PTD prevalence (41/932) does not transfer to infant ALL, where PTDs are rare and reciprocal translocations dominate. Orthogonal confirmation was possible in only 22 of the cases. The relevance here is mechanistic - an intragenic event is invisible to banding and to a break-apart probe - not epidemiological.

    Evidence that argues against this (1)
    • The entry describes the PTD as hard to detect and then reports a frequency from it — 41 of 932 — which is a lower bound from a cohort where confirmation required cross-checking NGS against MLPA and optical genome mapping. Centres without all three would have found fewer. The same caution applies to every historical KMT2A-PTD frequency in the literature. Journal of Molecular Diagnostics 2013
Connected to