Oncofusion

start / graph / target-phase2-dot1l-dossier

Targetstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q14-phase2-readiness

Phase-2 dossier: DOT1L - perfect chemistry, a completed paediatric trial, and a target that has already had its answer

also called DOT1L dossier · pinometostat readiness · EPZ-5676 phase-2 plan

DOT1L was the textbook case: a clean enzyme, a picomolar drug, and a paediatric trial that actually finished. It did not work well enough on its own, so the only honest phase-2 question left is whether it is useful in combination.

The same thing, for a scientist

DOT1L has everything a target is supposed to have: a solved co-crystal with the clinical compound (4HRA, DOT1L with EPZ-5676, 3.15 A), 193 ChEMBL activity rows at pChEMBL 6 or above with a best reported Kd of 0.01 nM, a completed Phase 1 dose-escalation and expansion study in paediatric relapsed/refractory KMT2A-rearranged leukaemia (NCT02141828), and a strong lineage-conditioned co-dependency with MEN1 (r 0.570 pooled, 0.544 within-tissue, 0.681 leukaemia-only). It is nonetheless the clearest single-agent disappointment in this field. Phase-2 readiness for DOT1L is therefore not a validation plan; it is a combination plan with a hard stopping rule, and the dossier is written to make restarting it expensive.

Go deeper
What structure exists

4HRA, crystal structure of DOT1L in complex with EPZ-5676, X-ray, 3.15 A, verified against RCSB this cycle 1. The catalytic SAM-binding site is fully characterised and the compound's binding mode is known. Nothing structural is missing.

What pocket is engaged

The SAM cofactor pocket of the DOT1L methyltransferase domain. This is an enzyme active site, not a protein-protein interface, which is why the chemistry was easy and why the target is not in the same class as menin or LEDGF. Critically, DOT1L is recruited to KMT2A fusion complexes indirectly, through AF9/ENL and AF10, and is not a direct KMT2A-binding partner - so inhibiting it does not disassemble the fusion complex, it removes one downstream output of it.

Essentiality and the normal-tissue picture

In this project's lineage-conditioned DepMap analysis, DOT1L retained 95% of its pooled correlation with MEN1 after within-tissue centring (0.570 to 0.544) and reached 0.681 within leukaemia lines 2 - a genuinely robust shared-process signal.

GTEx v8: expressed in all 54 tissues, 7.01 TPM in whole blood, minimum 3.87 TPM in brain amygdala, and a striking maximum of 111.7 TPM in testis 3. That testis enrichment is the tox flag that matters in a paediatric disease where survivors live for decades: gonadal DOT1L inhibition in a child is a fertility question nobody has asked. gnomAD pLI 1.0, LOEUF 0.288 4.

Chemical matter and clinical fate

ChEMBL target CHEMBL1795117: 193 activity rows at pChEMBL 6 or above, best reported Kd 0.01 nM (CHEMBL5764114) 5. The paediatric study NCT02141828, a Phase 1 dose escalation and expanded cohort of EPZ-5676 in children with relapsed/refractory KMT2A-rearranged leukaemia, is recorded as COMPLETED 6. Pharmacodynamic target engagement (H3K79me2 reduction) was demonstrated; clinical responses were not sufficient to carry the programme forward as a single agent.

This is the most instructive failure in the graph, because nothing was wrong with the chemistry. What was wrong was the assumption that removing one downstream mark would collapse a programme held up by several.

What would have to be true for a phase-2 programme

Only one thing would justify spending on DOT1L again: evidence that it delays the emergence of menin-inhibitor resistance, rather than that it adds depth of response. Depth of response was already measured and was not enough. Time to resistance was never measured.

The programme is therefore a barcoded in vitro resistance experiment, not a dose-response experiment, and the stopping rule is written against that endpoint. It also feeds directly into hyp-menin-plus-engager-orthogonal-pre-emption, which predicts that a second chromatin agent will not pre-empt resistance - meaning this dossier is designed to test a hypothesis this graph expects to fail.

Facts and figures
target genegene-dot1l
uniprotQ8TEK3
pocketSAM cofactor pocket of the DOT1L methyltransferase domain (enzyme active site, not a protein-protein interface)
recruitmentindirect - via AF9/ENL and AF10; DOT1L is not a direct KMT2A-binding partner, so inhibition removes an output rather than disassembling the complex
phase2 verdictDO NOT RESTART as a monotherapy programme. Restart only as the combination resistance experiment above, at low cost, and only after the LEDGF and menin programmes are running. On current evidence this is the candidate closest to being killed outright.
compute backendNone required beyond laptop-scale analysis; the work is wet-lab.
essentiality
lineage_conditioned_r_vs_…0.544lineage_conditioned_r_vs_MEN1: 0.5439pooled_r_vs_MEN10.570pooled_r_vs_MEN1: 0.5703leukemia_only_r_vs_MEN10.681leukemia_only_r_vs_MEN1: 0.6808fraction_retained0.954fraction_retained: 0.954
structures verified this cycle
[
  {
    "pdb": "4HRA",
    "title": "Crystal Structure of DOT1L in Complex with EPZ-5676",
    "method": "X-ray",
    "resolution_A": 3.15
  }
]
essentiality — the numbers behind the chart
{
  "lineage_conditioned_r_vs_MEN1": 0.5439,
  "pooled_r_vs_MEN1": 0.5703,
  "leukemia_only_r_vs_MEN1": 0.6808,
  "fraction_retained": 0.954
}
normal tissue gtex v8 tpm
{
  "whole_blood": 7.009,
  "min": {
    "tissue": "Brain_Amygdala",
    "tpm": 3.872
  },
  "max": {
    "tissue": "Testis",
    "tpm": 111.665
  },
  "n_tissues": 54
}
gnomad
{
  "pLI": 1,
  "loeuf": 0.288
}
chemical matter
{
  "chembl_target": "CHEMBL1795117",
  "rows_pchembl_ge_6": 193,
  "best": {
    "molecule": "CHEMBL5764114",
    "type": "Kd",
    "value_nM": 0.01,
    "pchembl": 11
  }
}
clinical history
{
  "paediatric_trial": "NCT02141828",
  "status": "COMPLETED",
  "phase": "PHASE1",
  "verdict": "pharmacodynamic target engagement achieved; single-agent clinical activity insufficient to advance"
}
go no go
{
  "G1_feasibility": {
    "question": "Can a barcoded time-to-resistance assay be built?",
    "GO": "A lentivirally barcoded KMT2A::AFF1 line maintains barcode diversity for 4 weeks under vehicle and loses it reproducibly under menin inhibitor alone",
    "STOP": "Barcode diversity collapses under vehicle - the assay cannot measure what it claims to"
  },
  "G2_biology": {
    "question": "Does DOT1L inhibition delay menin-inhibitor resistance?",
    "GO": "Time to outgrowth extended at least 2-fold versus menin inhibitor alone, replicated in two KMT2A-rearranged lines",
    "STOP": "Extension below 1.3-fold, or resistant outgrowth in the combination arm carries the same escape route as the single-agent arm",
    "ambiguous_zone": "1.3 to 2.0-fold",
    "tie_breaker": "Repeat in a primary infant KMT2A::AFF1 xenograft; result is binding"
  },
  "G3_window": {
    "question": "Gonadal and normal-progenitor cost",
    "GO": "No reduction in healthy-donor CD34+ CFU beyond 30% at combination-effective exposure",
    "STOP": "Any evidence of germ-cell toxicity in a juvenile animal model, given the 111.7 TPM testis expression and a survivor population that lives for decades"
  }
}
How this could bite us back
  • DOT1L is 16-fold enriched in testis over whole blood. In a disease whose survivors are children, gonadal toxicity is a decades-long liability that a short trial cannot detect.
  • DOT1L is recruited indirectly, so inhibiting it cannot disassemble the fusion complex; any benefit is a partial suppression of one output and is intrinsically escapable.
  • Adding a second chromatin-directed agent to a menin inhibitor may select for the same cell-state escape that the menin inhibitor alone selects for, in which case the combination buys nothing and costs toxicity - which is exactly what hyp-menin-plus-engager-orthogonal-pre-emption predicts.
Still open
  1. Was pinometostat's failure a target failure or a pharmacokinetic one - would a compound with better exposure have produced a different paediatric result?
  2. Does DOT1L inhibition change which resistance route a menin inhibitor selects, even if it does not delay resistance overall?
  3. Has anyone measured gonadal H3K79me2 in a juvenile animal exposed to a DOT1L inhibitor, given the testis expression?
Every source behind this entry (6)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    A co-crystal of DOT1L with the clinical compound EPZ-5676 exists (PDB 4HRA, X-ray, 3.15 A), verified by direct query of the RCSB data API this cycle.

    What this evidence cannot show

    3.15 A is modest for a structure-based design campaign; higher-resolution DOT1L structures exist with other ligands.

    Evidence that argues against this (1)
    • A 3.15 A co-crystal with the clinical compound is a good starting point and DOT1L's problem was never structural or potency-related. The compound reached patients, reduced H3K79 methylation demonstrably, and produced 2 complete remissions in 51 — so a dossier built on structure and affinity is measuring the axis that did not fail. RCSB PDB 2026
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    DOT1L's co-dependency with MEN1 survives lineage adjustment almost completely (pooled r 0.570, within-tissue 0.544, leukaemia-only 0.681) in this project's executed DepMap analysis.

    What this evidence cannot show

    54 leukaemia lines; DepMap release version unverified. A robust co-dependency is a statement about shared requirement in cell lines, and DOT1L's clinical result shows directly that such a signal does not guarantee single-agent benefit in patients.

    Evidence that argues against this (1)
    • The completed paediatric trial is the counter-evidence: the strongest lineage-robust co-dependency in this dataset after KMT2A itself still produced an inadequate single-agent clinical result. source 2026
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    DOT1L is expressed across all 54 GTEx v8 tissues with a pronounced testis maximum of 111.7 TPM against 7.01 TPM in whole blood and a minimum of 3.87 TPM in brain amygdala.

    What this evidence cannot show

    GTEx testis samples are adult. Whether prepubertal gonadal tissue shows the same enrichment is not addressed by this dataset, and that is precisely the population at risk.

    Evidence that argues against this (1)
    • GTEx TPM is BULK tissue RNA. It gives neither protein abundance nor cell-type resolution within a tissue, so a low whole-blood figure is compatible with high expression in the progenitor compartment that matters here, and a high figure in a bulk tissue does not localise the risk. The testis maximum is the classic pattern for a gene with a germline role, and testis expression is the single least informative tissue for predicting adult somatic toxicity. GTEx Portal 2026
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    DOT1L is strongly constrained against loss of function in humans (pLI 1.0, LOEUF 0.288).

    What this evidence cannot show

    Constraint is a germline, lifetime measure and did not predict the clinical tolerability of pinometostat, which was acceptable.

    Evidence that argues against this (1)
    • A gnomAD constraint metric is not a safety signal for a drug, and this beat uses it as one across every dossier. pLI and LOEUF measure intolerance to HETEROZYGOUS loss-of-function across a whole lifetime from conception, in a population sample. A drug gives partial, reversible occupancy in an adult for weeks. High constraint says the gene matters in development; it does not predict on-target toxicity of transient pharmacological inhibition, and this project has already had to make the same correction for germline MEN1 penetrance in the endocrine beat. gnomAD, Broad Institute 2026
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    ChEMBL holds 193 activity rows against DOT1L (CHEMBL1795117) at pChEMBL 6 or above, with a best reported affinity of Kd 0.01 nM - the deepest chemical matter of any non-menin target in this beat.

    What this evidence cannot show

    A 0.01 nM Kd in a biochemical assay for a slow-binding SAM-competitive inhibitor is not a cellular potency; pinometostat required long exposures for pharmacodynamic effect.

    Evidence that argues against this (1)
    • A ChEMBL activity-row count measures how much medicinal chemistry has been DONE, not how tractable the target is. The best reported affinity comes from the most optimised compound in whichever series was pursued, so it is a property of the investment rather than of the site, and comparing row counts across targets ranks attention rather than druggability. DOT1L's 193 rows reflect a completed industrial programme whose clinical answer is already known, which is why row count cannot rank it against an unexplored target. ChEMBL, EMBL-EBI 2026
  6. E6established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.trial_registry

    A Phase 1 dose-escalation and expanded-cohort study of EPZ-5676 (pinometostat) in paediatric patients with relapsed/refractory KMT2A-rearranged leukaemia was conducted and is recorded as completed.

    NCT02141828
    What this evidence cannot show

    Registry status was fetched this cycle and reflects the record, not the published outcome; the efficacy characterisation here is drawn from the field's consensus and is not read off the registry.

    Evidence that argues against this (1)
    • The paediatric phase 1 is the most relevant study in the dossier and its informative content is negative: dose escalation in relapsed paediatric leukaemia without a breakthrough efficacy signal. A go/no-go register should weight a completed paediatric trial above any preclinical metric in the same entry. RCSB PDB 2026
Connected to