Oncofusion

start / graph / strategy-collaborate-versus-compete

Groupmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q9-landscape

Collaborate or compete: an honest matrix, including where the lab would be duplicating a better-resourced effort

also called competitive assessment · who to partner with · where not to compete

Not everyone in this field is a potential partner, and some of what the lab might do is already being done better elsewhere. This entry says plainly which is which.

The same thing, for a scientist

A decision matrix over every group catalogued in the landscape beat, classifying each as natural collaborator, gatekeeper, competitor, or foundational-but-unavailable, with the specific asset each holds and the specific return offer the Iqbal Lab can make. It closes with the honest part: three areas where a small lab would be duplicating a far better-resourced effort and should not start.

Go deeper
Priority three, and why2 deeper

1. Interfant (Princess Maxima Center, Utrecht). The only global infant-ALL trial network, and the only holder of breakpoint-derived molecular data attached to protocol-standardised outcomes for 249 KMT2A-rearranged infants 1. They need new biology; the lab needs patients and annotated data. The ask costs them nothing - a retrospective analysis of data they already hold - and answers a twenty-year-old question. Highest value, lowest cost, and if it works it establishes the lab inside the only consortium that matters for this disease.

2. Frankfurt DCAL (Marschalek and Meyer). The world's largest nucleotide-resolution KMT2A breakpoint collection, 3,401 patients accumulated over two decades of diagnostic referral 2. Not replicable by anyone. They catalogue; they do not analyse the sequence for mechanism or prognosis. Purely computational collaboration, no competitive overlap, and it is the natural partner to the Interfant project rather than an alternative to it.

3. Thomas Milne (Oxford). The natural scientific collaborator. Owns Micro-Capture-C, the validated MLL-AF4 target gene set, and the PROM1/CD133 finding that became a CAR target 3. Runs no menin or chemistry programme, so overlap is near zero. The lab can offer interface predictions and a systematic fusion-surfaceome analysis; Milne can supply the only assay that can test whether disrupting a partner interface collapses the fusion's enhancer architecture.

Full matrix

Natural collaborators, low overlap:

  • Milne (Oxford) - enhancer architecture, PROM1, Micro-Capture-C
  • Interfant / Pieters / van der Sluis (Utrecht) - patients, outcomes, breakpoint-derived data
  • Marschalek and Meyer (Frankfurt) - breakpoint sequence at scale
  • Menendez and Bueno (Barcelona) - fetal and pluripotent human models, prenatal cohort methodology
  • Ernst (Colorado) - normal-function genetics, the therapeutic-window experiment
  • Bernt (Penn) - the DOT1L cautionary tale, epigenetic-mark rigour
  • Greaves (ICR) - aetiology framing only, and only on the narrow question of why infant KMT2A-r breaks his own model
  • The second ring (Bomken, Hagstrom-Andersson, Heidel, Zhu, Leutz, Inoue, Moore) - reciprocal, co-leadable, and where the non-menin dependencies are appearing

Gatekeepers, not collaborators:

  • Stein, Issa and the clinical menin investigator network - control trial access, but their programme is adult NPM1-mutant AML
  • Children's Oncology Group - controls North American infant biospecimens and, through AALL2121, the one genuinely infant-inclusive menin trial
  • The companies (Syndax, Kura, J&J, Sumitomo) - control the compounds and the trial designs

Competitors:

  • Armstrong and Krivtsov (Dana-Farber) - on menin mechanism and resistance, comprehensively
  • Grembecka and Cierpicki (Michigan) - complementary in capability, but a competitor on any interface they decide to pursue

Foundational but not available:

  • Cleary (Stanford) and Yokoyama (National Cancer Center Japan) - intellectual lineage; Yokoyama is the approachable half

Not a partner at all:

  • Biomea Fusion - its lead menin programme has moved to metabolic disease; both oncology trials terminated 4
Where the lab would be duplicating a better-resourced effort — do not start these

This is the section the lab will least want to read and most needs to.

1. Menin-inhibitor resistance mechanism. Dana-Farber, MSKCC and Syndax published the MEN1 interface-mutation mechanism, confirmed it in xenografts, and ran an unbiased base-editor screen to enumerate escape mutations before they appeared in patients 5. They have the patient samples from the registrational trial. A small lab entering here arrives second with worse material. The only defensible slice is the question they have not answered - how KMT2A-rearranged patients relapse without a MEN1 mutation - and even that is exposed to being scooped by whoever holds the next relapse cohort.

2. Novel menin-inhibitor chemistry. Five companies have clinical-stage assets and Michigan has the academic structural franchise. Making a sixth menin binder is a waste of a small lab's decade.

3. Building a KMT2A-r PDX bank. Multi-year, expensive, and Dana-Farber with Children's Cancer Institute (UNSW) already has one with menin-inhibitor response data 6. Borrow, do not build.

The honest generalisation: the lab's comparative advantage is analysis, synthesis and hypothesis generation across data other people hold, plus the willingness to ask questions that are not commercially interesting. It is not wet-lab throughput, not chemistry, and not clinical access. Every strategy in this graph should be built on the first and route around the other three.

Facts and figures
groupDecision matrix across the full landscape
institutionMultiple
focusClassifying every catalogued group as collaborator, gatekeeper, competitor or unavailable, and identifying where the lab should not compete
partnering angleApproach Interfant, Frankfurt and Milne first, in that order, each with a specific computational return offer. Treat the clinical menin network, COG and the companies as gatekeepers to be engaged later and narrowly. Do not start work on menin-inhibitor resistance mechanism, novel menin chemistry, or building a PDX bank.
key papers
A Menin-MLL Inhibitor Ind…2,019A Menin-MLL Inhibitor Induces Specific Chromatin Changes and Eradicates Disease in Models of MLL-Rearranged Leukemia: 2019MEN1 mutations mediate cl…2,023MEN1 mutations mediate clinical resistance to menin inhibition: 2023MLL-AF4 cooperates with P…2,023MLL-AF4 cooperates with PAF1 and FACT to drive high-density enhancer interactions in leukemia: 2023
priority three
collab-pieters-interfant1.000collab-pieters-interfant: 1collab-marschalek-meyer-f…2.000collab-marschalek-meyer-frankfurt: 2collab-milne-oxford3.000collab-milne-oxford: 3
key papers — the numbers behind the chart
[
  {
    "title": "MEN1 mutations mediate clinical resistance to menin inhibition",
    "year": 2023,
    "venue": "Nature",
    "pmid": "36922589",
    "doi": "10.1038/s41586-023-05755-9"
  },
  {
    "title": "A Menin-MLL Inhibitor Induces Specific Chromatin Changes and Eradicates Disease in Models of MLL-Rearranged Leukemia",
    "year": 2019,
    "venue": "Cancer Cell",
    "pmid": "31821784",
    "doi": "10.1016/j.ccell.2019.11.001"
  },
  {
    "title": "MLL-AF4 cooperates with PAF1 and FACT to drive high-density enhancer interactions in leukemia",
    "year": 2023,
    "venue": "Nature Communications",
    "pmid": "37626123",
    "doi": "10.1038/s41467-023-40981-9"
  }
]
priority three — the numbers behind the chart
[
  {
    "rank": 1,
    "partner": "collab-pieters-interfant",
    "why": "only holder of breakpoint-derived molecular data linked to protocol-quality infant outcomes; the ask costs them nothing",
    "ask": "retrospective analysis of 249 archived KMT2A breakpoints against existing outcome data",
    "offer": "full statistical design, pre-registration and analysis at no cost to them"
  },
  {
    "rank": 2,
    "partner": "collab-marschalek-meyer-frankfurt",
    "why": "irreplaceable 3,401-patient breakpoint dataset that they catalogue but do not analyse for mechanism",
    "ask": "computational reanalysis of existing breakpoint sequences (domain content, junction features, reciprocal status)",
    "offer": "the analysis layer they have never applied to their own data, and a bridge to Interfant outcomes"
  },
  {
    "rank": 3,
    "partner": "collab-milne-oxford",
    "why": "the only group whose assay can test whether interface disruption collapses fusion enhancer architecture; near-zero competitive overlap",
    "ask": "Micro-Capture-C readout of a predicted interface perturbation",
    "offer": "interface hot-spot predictions and a systematic fusion-surfaceome analysis beyond PROM1, directly useful to their CAR programme"
  }
]
do not start
[
  {
    "area": "menin-inhibitor resistance mechanism",
    "held_by": "collab-armstrong-krivtsov-dfci and collab-stein-issa-clinical-menin",
    "why": "MEN1 interface mutations published, xenograft-confirmed, base-editor-screened; they hold the registrational trial's relapse samples",
    "narrow_exception": "relapse WITHOUT a MEN1 mutation, which is unexplained"
  },
  {
    "area": "novel menin-inhibitor chemistry",
    "held_by": "five clinical-stage companies plus collab-grembecka-cierpicki-michigan",
    "why": "a sixth menin binder adds nothing and costs a decade",
    "narrow_exception": "none"
  },
  {
    "area": "building a KMT2A-r PDX bank",
    "held_by": "collab-armstrong-krivtsov-dfci with Children's Cancer Institute UNSW",
    "why": "multi-year, expensive, already exists with drug-response annotation",
    "narrow_exception": "borrow via the UNSW arm rather than build"
  }
]
How this could bite us back
  • Classifying Armstrong as a competitor is correct today but self-fulfilling if it stops the lab from ever approaching them; the cost of a declined email is zero and the cost of avoidance is a permanently closed door.
  • A strategy built on three collaborations has three single points of failure, all outside the lab's control.
  • Declaring areas off-limits risks the lab abandoning ground it could have won if the incumbent's programme stalls, which happens routinely.
  • Everything in this matrix is inferred from publication and registry records. Unpublished work, funded-but-unstarted programmes and personnel moves are invisible, and any of them could invert a classification overnight.
Still open
  1. Is the DFCI/UNSW PDX bank available to outside collaborators, and is the Sydney arm a more accessible route than Boston?
  2. How do KMT2A-rearranged patients relapse on a menin inhibitor when they do not acquire a MEN1 mutation? This is the one slice of the resistance question that remains open, and whoever gets the next relapse cohort will close it.
  3. If all three priority collaborations are declined, what is the lab's plan? A strategy with no fallback for its top three asks is not a strategy.
  4. Should the lab approach the second ring (Bomken, Hagstrom-Andersson, Heidel) FIRST to build a publication record, on the grounds that Interfant and Frankfurt are more likely to say yes to a group with a recent relevant paper?
Every source behind this entry (6)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Interfant-06 holds MRD data derived from patient-specific DNA-based KMT2A assays for 249 infants together with protocol-standardised outcome data, and the consortium spans 18 national and international study groups.

    PMID 33405950DOI
    What this evidence cannot show

    Retention of the underlying breakpoint sequences is inferred, not stated. Consortium governance across 18 groups is a real and underestimated obstacle.

    Evidence that argues against this (1)
    • Qualifies priority one. Interfant's successor protocol is open and it is not a menin study. Interfant-21 (NCT05327894) has been RECRUITING since December 2022 with an estimated 160 infants under one year with KMT2A-rearranged ALL or MPAL, sponsored by the Princess Maxima Center with the University of Milano-Bicocca and Amgen Europe B.V.; its only registered interventions are blinatumomab, its design is phase 3, NON_RANDOMIZED and sequential, with risk allocation by age under 6 months, white-cell count and prednisone response, and a primary endpoint of event-free survival. Three consequences for the ranking in this entry. (1) A lab whose central thesis is the menin interface would be entering its highest-priority collaboration with the one infant network not testing menin inhibition; the North American routes (COG AALL2121, NCT05761171, revumenib; TINI 2, NCT05848687, ziftomenib) carry that question instead. (2) The proposed retrospective - breakpoint position against protocol-standardised outcome - is a proposal about a closed cohort (Interfant-06) offered to a consortium whose attention is on an open one, which is a harder sell than the entry assumes. (3) Because Interfant-21 is non-randomised, its blinatumomab effect will again be read against historical comparators, so the methodological weakness noted for the 30-infant study is structural rather than transitional. None of this removes Interfant from the top of the list for patients and outcome data; it changes what should be asked for and when. Caveat: registry records summarise rather than reproduce a protocol; correlative biology and molecular annotation can exist without appearing as registered outcome measures. Read: full registry record, retrieved 2026-09-18. Princess Maxima Center for Pediatric Oncology; University of Milano Bicocca; Amgen Europe B.V. 2026
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The Frankfurt DCAL series holds nucleotide-resolution KMT2A breakpoints for 3,401 patients analysed 2003-2022, with 107 in-frame fusions catalogued, accumulated through diagnostic referral rather than through a research cohort.

    The KMT2A recombinome of acute leukemias in 2023. · Meyer C, ... Marschalek R · Leukemia 2023
    PMID 37019990DOI
    What this evidence cannot show

    Referral bias; no published data-sharing policy; operational status of the centre in 2026 unverified.

    Evidence that argues against this (1)
    • Refutes this evidence item's characterisation of the Frankfurt group. E2 says of Marschalek and Meyer: 'They catalogue; they do not analyse the sequence for mechanism or prognosis.' Both halves are wrong as statements about the group. On prognosis: Emerenciano, Meyer, Mansur, Marschalek and Pombo-de-Oliveira published exactly that analysis in 2013, correlating genomic breakpoint location with outcome in childhood leukaemia at or under 24 months and reporting that poor-outcome KMT2A-rearranged patients preferentially carried breakpoints in MLL intron 11. On mechanism: Marschalek is a co-author on the negative human-HSPC test of AF4-MLL leukaemogenicity (Prieto et al., Oncotarget 2017, PMID 29137234, DOI 10.18632/oncotarget.19567) and on the positive MLL-AF4 plus AF4-MLL cooperation study in human embryonic stem cell haemato-endothelial differentiation (Bueno et al., Haematologica 2019, PMID 30679325, DOI 10.3324/haematol.2018.202044). The strategic consequence is real and is not favourable. The pitch this entry proposes for priority two - 'purely computational collaboration, no competitive overlap' because they do not analyse - rests on a false premise. Frankfurt has already asked the prognostic question and holds the mechanistic interest too, so the correct framing is a confirmation-and-extension proposal to a group that has been there, with the 2013 paper cited in the first paragraph, not a gap-filling proposal. Caveat: three papers do not establish a research programme, and it remains true that the 3,401-patient recombinome series is published as a catalogue; the error is in the claim about the group, not about that particular paper. Read: abstract only. Emerenciano M et al. 2013
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Milne's group owns the PROM1/CD133 fusion-target finding, the enhancer-architecture methodology (Micro-Capture-C showing MLL-AF4-driven high-density enhancer-promoter interactions enriched for PAF1C and FACT), and runs no competing menin or chemistry programme.

    PMID 37626123DOI
    What this evidence cannot show

    'Runs no competing programme' is an assessment from publication record, not a statement of intent; group priorities change and unpublished work is invisible.

    Evidence that argues against this (1)
    • "Owns" is a claim about priority and priority is contestable: papers have many authors and several institutions, Micro-Capture-C is a published method that other groups can run, and PROM1 as a fusion target has been reported by more than one group. Running no competing programme is also a fact with a short half-life. eLife 2021
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.trial_registry

    Biomea Fusion's menin inhibitor icovamenib has moved away from oncology: both of its oncology trials are terminated and its active and completed studies are metabolic, so it is not a partner for a leukaemia programme.

    ClinicalTrials.gov sponsor records for Biomea Fusion, retrieved via the ClinicalTrials.gov API v2 · US National Library of Medicine; Biomea Fusion · ClinicalTrials.gov 2026
    What this evidence cannot show

    Verified by a worker agent in this same survey cycle from registry records and the company's own pipeline page. A company's own pipeline page is a claim by that party, not an independent fact; registry status can lag corporate decisions in either direction, and a company can re-enter an indication. See collab-biomea-fusion for the primary verification.

    Evidence that argues against this (1)
    • Corrects a supporting fact. This item says Biomea 'has moved away from oncology' and cites a ClinicalTrials.gov sponsor search. The conclusion is right and the record is now more specific than the entry allows: the COVALENT-101 registry record (NCT05153330) carries a populated WhyStopped field reading 'Biomea Fusion, Inc., is no longer pursuing oncology indications for BMF-219. No safety concerns or efficacy observations led to this study closure.' That matters for the competitive matrix in a way the entry does not draw out: the covalent menin inhibitor left leukaemia for portfolio reasons, not because covalent menin engagement failed, so its absence is not evidence about the target and the covalent modality remains an untested differentiator rather than a discredited one. Caveat: WhyStopped is sponsor-submitted, unaudited free text and a sponsor has an obvious interest in disclaiming safety and efficacy problems. Read: registry record only, retrieved 2026-09-18. Biomea Fusion et al. 2026
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The Dana-Farber/MSKCC/Syndax axis identified MEN1 interface mutations as the mechanism of acquired clinical resistance to menin inhibition, confirmed it in xenograft models, and ran an unbiased base-editor screen, occupying the resistance-mechanism question comprehensively.

    MEN1 mutations mediate clinical resistance to menin inhibition · Perner F, Stein EM, Wenge DV, et al., Levine RL, Armstrong SA, Cai SF · Nature 2023
    PMID 36922589DOI
    What this evidence cannot show

    They occupied the MEN1-mutation route specifically. Non-genetic and non-MEN1 resistance is not closed by this paper, which is why one narrow slice remains open.

    Evidence that argues against this (1)
    • Sharpens the competitor classification with a documented interest the entry does not record. The competing-interests statement of the cited resistance paper reads, for S.A.A.: 'has been a consultant and/or shareholder for Vitae/Allergan Pharmaceuticals, Neomorph, Inc., Imago Biosciences, Cyteir Therapeutics, C4 Therapeutics, and Accent Therapeutics. S.A.A. has received research support from Janssen, and Syndax and is an inventor on a patent related to Menin Inhibition WO/2017/132398A1.' A 2025 Cancer Research disclosure from the same investigator adds 'grants from Syndax Pharmaceuticals and Janssen Biotech' and 'a patent for "Menin inhibition in NPM1 AML" issued' (PMID 40163352, DOI 10.1158/0008-5472.CAN-24-1136). So the group at the top of this entry's competitor list is not only ahead on the resistance question; it holds issued patent claims on menin inhibition and current research funding from two of the four clinical sponsors. That is a stronger reason to route around menin than 'they occupy the question comprehensively', and it argues that any chemical matter a collaboration generated at this interface should be freedom-to-operate reviewed before a single compound is made. Caveats: declared interests are disclosures, not conduct findings, and a patent citation in a disclosure statement is not a claim analysis - the scope of WO/2017/132398A1 was not examined here. Read: competing-interests sections only; abstracts read, not full texts. Perner F et al. 2023
  6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    A KMT2A-rearranged ALL and AML patient-derived xenograft bank with menin-inhibitor response data already exists, built jointly by Dana-Farber and the Children's Cancer Institute at UNSW Sydney, with some ALL xenografts remaining disease-free for over a year after treatment.

    A Menin-MLL Inhibitor Induces Specific Chromatin Changes and Eradicates Disease in Models of MLL-Rearranged Leukemia · Krivtsov AV, Evans K, Gadrey JY, et al., Lock RB, Armstrong SA · Cancer Cell 2019
    PMID 31821784DOI
    What this evidence cannot show

    Whether that bank is available to outside collaborators, and on what terms, was not verified. The UNSW arm (Richard Lock, Children's Cancer Institute) may be a more accessible route to PDX work than the DFCI arm and is worth approaching separately.

    Evidence that argues against this (1)
    • A bank existing is not a bank being available. PDX access runs through material transfer agreements, institutional review and the originating investigators' own plans, and xenografts drift on passage — clonal composition and drug response change with serial transplantation, so menin-response data attached to an early passage may not hold for the vial that arrives. Nature 2023
Connected to