Rolf Marschalek & Claus Meyer (DCAL, Goethe-University Frankfurt) — the world's KMT2A breakpoint dataset
also called DCAL · Diagnostic Center of Acute Leukemia · Frankfurt recombinome group · Marschalek Lab · Institute of Pharmaceutical Biology Goethe-University
A single laboratory in Frankfurt has sequenced the exact DNA break point in more KMT2A-rearranged leukemia patients than everyone else combined, because for twenty years the world has been sending them samples for diagnosis. If you want to ask any question about where the break happens, you have to go through them.
The same thing, for a scientist
The Diagnostic Center of Acute Leukemia (DCAL) at the Institute of Pharmaceutical Biology, Goethe-University Frankfurt, run by Rolf Marschalek with Claus Meyer as lead analyst, has used long-distance inverse PCR to characterise genomic KMT2A breakpoints and translocation partner genes as a diagnostic service since the early 2000s. The successive 'MLL/KMT2A recombinome' papers report 1,590 samples (2013), 2,345 patients (2017) and 3,401 patients analysed 2003-2022 (2023), cumulatively defining 107 in-frame KMT2A gene fusions plus out-of-frame, partnerless and partial-tandem-duplication classes. The 2017 paper reports an age-dependent breakpoint shift within the KMT2A breakpoint cluster region.
What they own that nobody else can generate2 deeper
Genomic breakpoint sequences at nucleotide resolution for thousands of patients, accumulated over two decades. This is not a dataset that can be recreated. It exists because DCAL operated as a diagnostic referral service: hospitals across Europe and beyond sent biopsy material to Frankfurt to have the rearrangement characterised, and the sequences accumulated as a by-product of clinical service 123.
The scale: 1,590 MLL-rearranged biopsy samples by 2013 3, 2,345 patients by 2017 2, 3,401 patients analysed between 2003 and 2022 by the 2023 update 1. Including literature, 107 in-frame KMT2A fusions are now defined, with a further 16 out-of-frame fusions, 18 patients with no partner gene fused to 5'-KMT2A, two with a 5'-KMT2A deletion 1. Nine specific gene fusions account for more than 90% of all illegitimate KMT2A recombinations 2.
The age-dependent breakpoint shift. The 2017 paper reports that the position of the break inside the KMT2A breakpoint cluster region shifts with age at diagnosis 2. That single observation is the strongest published link between the mechanism of fusion formation and the clinical phenotype, and it is theirs.
A mechanistic position the field does not fully share. Marschalek argues that the reciprocal AF4-MLL product and the wild-type AF4 multiprotein complex are causally important, not incidental 4. Most of the field concentrates on the direct fusion. Whoever is right, the reciprocal-product question cannot be studied without breakpoint-resolved patient material.
Why this is the single highest-value collaboration on the list
The Iqbal Lab's most distinctive assets are computational and hypothesis-generating. The Frankfurt dataset is the one resource in the world where those assets convert directly into a publishable result with no wet-lab spend.
Concrete proposals, in ascending cost:
- Breakpoint position versus predicted retained/lost domain content, computed across all 3,401 cases. The recombinome papers report breakpoint distributions but the domain-level consequence per patient has not been systematically tabulated.
- Breakpoint sequence features versus formation mechanism (microhomology length, Alu/repeat content, topoisomerase-II consensus, R-loop propensity), which is pure sequence analysis on data they already hold.
- Breakpoint position as a prognostic variable, which requires linking their sequences to outcome data held by the trial consortia. That is the bridge project between Frankfurt and Interfant.
This is a collaborator, not a competitor. DCAL is a diagnostic and cataloguing operation; it does not run a target-discovery or therapeutics programme that would collide with the lab's.
What is not verified here
Whether the raw breakpoint sequences are deposited anywhere accessible, what the data-sharing terms are, whether DCAL is still operating as a referral service in 2026, and who currently leads it were not verified in this pass. The recombinome papers themselves are verified; the operational status of the centre is not. Treat any statement about current sample flow as an assumption to check before writing to them.
| group | Rolf Marschalek (head) and Claus Meyer, Diagnostic Center of Acute Leukemia (DCAL) |
|---|---|
| institution | Institute of Pharmaceutical Biology / DCAL, Goethe-University, Frankfurt am Main, Germany (affiliation as printed on PubMed-indexed papers through 2023) |
| focus | Nucleotide-resolution genomic breakpoint characterisation of KMT2A/MLL rearrangements by long-distance inverse PCR; translocation partner gene catalogue; reciprocal fusion products; age-dependent breakpoint distribution |
| partnering angle | The single highest-value data collaboration available. Propose a purely computational reanalysis of breakpoint sequences they already hold: (1) per-patient retained/lost domain content derived from breakpoint position, (2) sequence-feature analysis of formation mechanism (microhomology, repeat content, TOP2 consensus, R-loop propensity), (3) with Interfant, breakpoint position as a prognostic variable. Zero wet-lab cost to us, zero new sample collection for them, and a result neither party can produce alone. |
| relationship | natural collaborator, no overlap with our programme |
key papers — the numbers behind the chart
[
{
"title": "The KMT2A recombinome of acute leukemias in 2023",
"year": 2023,
"venue": "Leukemia",
"pmid": "37019990",
"doi": "10.1038/s41375-023-01877-1"
},
{
"title": "The MLL recombinome of acute leukemias in 2017",
"year": 2017,
"venue": "Leukemia",
"pmid": "28701730",
"doi": "10.1038/leu.2017.213"
},
{
"title": "The MLL recombinome of acute leukemias in 2013",
"year": 2013,
"venue": "Leukemia",
"pmid": "23628958",
"doi": "10.1038/leu.2013.135"
},
{
"title": "MLL leukemia and future treatment strategies",
"year": 2015,
"venue": "Arch Pharm",
"pmid": "25740345",
"doi": "10.1002/ardp.201400449"
}
]dataset size — the numbers behind the chart
{
"patients_2013": 1590,
"patients_2017": 2345,
"patients_2003_2022": 3401,
"in_frame_fusions_catalogued": 107,
"partner_genes_molecularly_characterised_2017": 94
}leverage they hold
[ "largest nucleotide-resolution KMT2A breakpoint collection in the world", "two decades of diagnostic referral flow from European centres", "long-distance inverse PCR assay expertise", "reciprocal-fusion characterisation nobody else has at scale" ]
not verified
[ "whether raw breakpoint sequences are publicly deposited and under what terms", "whether DCAL is still operating as a referral diagnostic service in 2026", "current leadership and funding status of the centre" ]
- A referral series is not a population series; any prognostic signal found in it may be a referral-pattern artefact rather than biology.
- Dependence on a single laboratory for the world's breakpoint data is itself a systemic risk: if DCAL closes or the PI retires, the resource may become inaccessible.
- If breakpoint position turns out to change retained domain content in a meaningful fraction of patients, a single menin-interface drug will not fit all KMT2A-r patients, which complicates the platform argument rather than supporting it.
- Does breakpoint position within the KMT2A breakpoint cluster region predict outcome once age is controlled for? The dataset to answer this exists in Frankfurt; the outcome data exists in the trial consortia; nobody has joined them.
- What fraction of the 3,401 cases have breakpoints that alter the retained menin-binding, CXXC and AT-hook content of the fusion? If any do, they are a natural experiment testing whether the menin interface is truly obligate.
- Are the 18 patients with no partner gene fused to 5'-KMT2A a real biological class or an assay artefact, and do they respond to menin inhibition?
- Are the raw breakpoint sequences deposited anywhere, or does every question require a collaboration agreement with one German laboratory?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The Frankfurt DCAL group reported genomic KMT2A breakpoint and translocation-partner-gene data from 3,401 acute leukaemia patients analysed between 2003 and 2022, with a cumulative total of 107 in-frame KMT2A gene fusions identified, plus 16 out-of-frame rearrangements, 18 patients with no partner gene fused to 5'-KMT2A and 2 with a 5'-KMT2A deletion.
The KMT2A recombinome of acute leukemias in 2023. · Meyer C, ... Marschalek R · Leukemia 2023What this evidence cannot show
This is a referral cohort, not a population cohort: samples arrive because a clinician suspected a KMT2A rearrangement and sought molecular characterisation, so partner-gene frequencies are subject to referral and geographic bias and cannot be read as true population frequencies. The series is also not linked to systematic outcome data in the published form.
Evidence that argues against this (1)
- The series is a diagnostic REFERRAL cohort, so every frequency it reports is conditioned on who was sent for KMT2A analysis and with which assay. This graph has already had to correct partner percentages read across that table as if the rows were ascertained alike: spliced fusions need RNA to see, cryptic partners need molecular work, and visible reciprocal translocations are over-represented. Leukemia 2023
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The 2017 recombinome reported 2,345 patients, 135 different MLL rearrangements with 94 partner genes characterised at the molecular level, 35 of them recurrent, with nine specific gene fusions accounting for more than 90% of illegitimate MLL recombinations, and an age-dependent shift in breakpoint position.
The MLL recombinome of acute leukemias in 2017 · Meyer C, ... Marschalek R · Leukemia 2017What this evidence cannot show
Same referral-bias limitation. The age-dependent breakpoint shift is a descriptive association in a heterogeneous referral series and has not, as far as this pass found, been independently replicated in a population-based cohort with matched outcome data.
Evidence that argues against this (1)
- The 2017 and 2023 editions are not independent observations. The later series subsumes the earlier patients, so quoting both as separate evidence double-counts a single accumulating cohort, and apparent stability of a proportion across editions is partly arithmetic. Leukemia 2018
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The method behind the series is long-distance inverse PCR applied to individual patient biopsy samples; the 2013 paper characterised 1,590 MLL-rearranged samples with precise breakpoint localisation within MLL and within the partner gene.
The MLL recombinome of acute leukemias in 2013 · Meyer C, ... Marschalek R · Leukemia 2013What this evidence cannot show
Long-distance inverse PCR has assay-specific blind spots: breakpoints falling outside the amplified region, complex three-way rearrangements and insertions can be missed or mis-assigned, so the catalogue is a lower bound on rearrangement diversity.
Evidence that argues against this (1)
- Long-distance inverse PCR is the method that built the series and it also sets its blind spot: breakpoints outside the amplifiable window are missed, which is exactly why a second cluster region took until 2019 to be described and why 21 of 93 KMT2A-rearranged patients had breakpoints two PCR methods could not find. Leukemia 2013
- E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.review
Marschalek argues that MLL fusion proteins retain binding to MEN1 and LEDGF to target promoters, and that the most frequent MLL fusions recruit the wild-type AF4 multiprotein complex; he also emphasises reciprocal MLL fusions (about 120 characterised) as potentially causal rather than incidental.
MLL leukemia and future treatment strategies · Marschalek R · Archiv der Pharmazie 2015What this evidence cannot show
A single-author review, so this is the group's interpretive position rather than independent evidence. The causal importance of the reciprocal AF4-MLL product is genuinely contested; other groups regard it as dispensable. Cite this as a stated position, not as an established fact.
Evidence that argues against this (1)
- A Menendez-group study co-authored with Marschalek found that expressing either MLL-AF4 or the reciprocal AF4-MLL in HEK293 cells did not alter DNA double-strand-break repair kinetics or NHEJ/HR efficiency, arguing against one proposed mechanism by which the reciprocal product would matter. Castano J et al. 2016
Source-concentration counter (Meyer 2023, PMID 37019990): the modern reference distribution of KMT2A breakpoints and partners is a single referral-based diagnostic centre (3401 patients, 2003-2022); no independent cohort has published a partner-stratified KMT2A intron distribution
hold the largest breakpoint dataset in the world
who holds the samples and data we could not generate ourselves
Counter-evidence (Emerenciano 2013, PMID 23432364; Prieto 2017, PMID 29137234; Bueno 2019, PMID 30679325): the claim that Frankfurt 'catalogue but do not analyse for mechanism or prognosis' is false -- they have published both, so the priority-two pitch rests on a false premise
- related to →KMT2A intron 9 / exon 9 breakpoints (BCR subregion A): the HOXA-high, older-patient break
nucleotide-resolution breakpoint characterisation by long-distance inverse PCR across 3,401 patients
source of the age-dependent breakpoint shift within the KMT2A breakpoint cluster region
- supports →Do different breaks give different biology? A verdict on partner identity vs breakpoint position
the catalogue and the age-dependent shift are the empirical basis for the claim that different breaks may mean different biology
holds the reference breakpoint catalogue needed to pre-specify breakpoint classes before testing them against outcome
- related to →KMT2A::MLLT10 (MLL-AF10), t(10;11)(p12;q23)
holds the spliced (reciprocal-free) versus reciprocal-expressing KMT2A::MLLT10 natural experiment
the irreplaceable 3,401-patient nucleotide-resolution breakpoint collection
priority collaboration 2
co-published, including a negative result on MLL-AF4 and AF4-MLL effects on DNA repair
THE SERIES' SOURCE: the concentration of this graph's partner-frequency evidence in one reference laboratory is what makes the denominator question load-bearing.
breakpoint-typed t(11;19) cases needed for arm B live in the recombinome cohorts