No menin-inhibitor trial measures homocysteine or any amino-acid endpoint across 57 records - while a recruiting infant-ALL trial already gives PEG asparaginase and ziftomenib together
also called homocysteine in no menin trial · CBS thromboembolism asparaginase · TINI 2 ziftomenib PEG asparaginase · NCT05848687 · 57 menin-inhibitor trials
This project found that a menin inhibitor strongly lowers the gene for an enzyme whose inherited loss causes dangerous blood clots, and it proposed combining that drug with asparaginase, which also causes clots. It turns out the combination is already being given to infants in an open trial - and the simple blood test that would detect the problem is named in none of the 57 trials of this drug class.
The same thing, for a scientist
Q109 measured **CBS falling 4.07 log2** under menin inhibition - the largest fall in its serine/one-carbon panel. CBS is cystathionine beta-synthase, and germline CBS deficiency causes classical homocystinuria whose cardinal complication is **thromboembolism**. **Asparaginase**, which this graph's own `hyp-menin-inhibition-collapses-the-serine-asns-programme-and-should-resensitise-kmt2a-r-all-to-asparaginase` proposes combining with a menin inhibitor, carries **thrombosis among its dose-limiting toxicities**. Q109 called plasma homocysteine in a menin-inhibitor trial *"a routine assay nobody has run"*. This asks the registry, and asks a second question it did not. **THE ANSWER, over 57 distinct trials of nine menin-inhibitor names.** Homocysteine appears as a named outcome measure in **0**. Cystathionine **0**. Any amino-acid or metabolomic endpoint **0**. Meanwhile **QTc appears in 2** — so these protocols **do** add analyte-level endpoints when a liability is recognised, which is what makes the asymmetry an argument rather than a shrug. **The controls caught a false result before it was believed.** The first version of this scan reported zero for **all fifteen** tokens, including "amino acid" — and a zero across an entire token list is instrument failure, which this project has a standing rule for. The control tokens prove the text is loaded: **adverse 45/57, dose 34/57, survival 30/57, response 28/57, safety 21/57**, across **113,991 characters** of outcome text with only **2 of 57** trials carrying an empty outcome blob. The script now asserts that at least one control fires before any zero is reported. **AND THE COMBINATION IS ALREADY IN PATIENTS — INFANTS.** Three trials mention asparaginase and all three co-administer a preparation of it with a menin inhibitor: - **NCT05848687 — "TINI 2: Total Therapy for Infants With Acute Lymphoblastic Leukemia II"**, **RECRUITING**, CHILD, and its intervention list contains **PEG asparaginase** and **Ziftomenib**. This is the combination this graph's standing hypothesis proposes, in the index population, now. - **NCT05761171** — revumenib with chemotherapy including **Calaspargase Pegol**, minimum age **1 Month**, relapsed/refractory ALL and AML. - **NCT07636564** — blinatumomab ± revumenib with **Calaspargase Pegol** in B-ALL. Of the 57, **17 mention lymphoblastic leukaemia**, **44 mention KMT2A or MLL-rearranged**, and **25 include the CHILD age group**. **The four "thrombo" outcome hits are incidental and I read all of them**: a post-essential-thrombocythemia condition name in a myelofibrosis trial, thrombocytopenia inside DLT boilerplate in two trials, and a healthy-volunteer food-effect study measuring a routine coagulation panel. **None is an ALL trial and none is a thrombosis endpoint.**
| script | analyses/homocysteine-is-measured-in-no-menin-inhibitor-trial/run.py |
|---|---|
| results | analyses/homocysteine-is-measured-in-no-menin-inhibitor-trial/results.json |
| status | run |
| run date | 2026-09-22 |
| source | ClinicalTrials.gov API v2, query.intr over revumenib, SNDX-5613, ziftomenib, KO-539, bleximenib, JNJ-75276617, DSP-5336, enzomenib and BMF-219, deduplicated by NCT id |
| n distinct trials | 57 |
| the thrombo hits read individually | four outcome hits, all incidental: a post-essential-thrombocythemia condition name in a myelofibrosis trial, thrombocytopenia inside DLT boilerplate in two trials, and a healthy-volunteer food-effect study measuring fibrinogen, prothrombin time and INR. None is an ALL trial; none is a thrombosis endpoint. |
| what this is NOT | evidence of harm. It is a gap in measurement beside a co-exposure that is already happening, which is a reason to measure and not a reason to stop. |
| absence class | never-measured |
the answer
{
"homocysteine as a named outcome measure": 0,
"cystathionine": 0,
"amino acid": 0,
"metabolomic": 0,
"serine": 0,
"folate": 0,
"methionine": 0,
"QTc": 2,
"the_asymmetry_is_the_argument": "the recognised liability gets a specific lab endpoint; the unrecognised one does not. These protocols are capable of itemising an analyte and do so when the risk is known."
}positive controls that make the zeros believable
{
"adverse": "45/57",
"dose": "34/57",
"survival": "30/57",
"response": "28/57",
"safety": "21/57",
"remission": "18/57",
"total_outcome_text_characters": 113991,
"trials_with_an_empty_outcome_blob": 2,
"why_this_block_exists": "the first version of this scan reported ZERO for all fifteen tokens including 'amino acid'. A zero across an entire token list is instrument failure and this project has a standing rule for it. The script now ASSERTS that at least one control fires before any zero is reported."
}the combination is already being given
{
"NCT05848687": {
"title": "TINI 2: Total Therapy for Infants With Acute Lymphoblastic Leukemia II",
"status": "RECRUITING",
"ages": "CHILD",
"asparaginase_preparation_listed_as_an_intervention": "PEG asparaginase",
"menin_inhibitor": "Ziftomenib",
"why_it_matters": "this is the menin-inhibitor-plus-asparaginase combination this graph's standing hypothesis proposes, in infants with ALL, currently recruiting"
},
"NCT05761171": {
"title": "A Study of Revumenib in Combination With Chemotherapy for Patients Diagnosed With Relapsed or Refractory Leukemia",
"status": "ACTIVE_NOT_RECRUITING",
"minimum_age": "1 Month",
"asparaginase_preparation": "Calaspargase Pegol"
},
"NCT07636564": {
"title": "Testing Blinatumomab With or Without Revumenib in Patients With B-cell Acute Lymphoblastic Leukemia With a Genetic Change Requiring More Treatment",
"status": "NOT_YET_RECRUITING",
"asparaginase_preparation": "Calaspargase Pegol"
}
}population exposure — the numbers behind the chart
{
"trials mentioning lymphoblastic": 17,
"trials mentioning KMT2A or MLL-rearranged": 44,
"trials including the CHILD age group": 25,
"of": 57
}- THIS IS NOT EVIDENCE OF HARM AND MUST NOT BE PRESENTED AS SUCH. It is a measurement gap sitting beside a co-exposure that is already happening. No clinical signal of hyperhomocysteinaemia on a menin inhibitor has been reported, and the entire mechanistic chain rests on one transcript measurement in a cell line.
- Nothing here is a reason to withhold or delay a menin inhibitor, or to alter an asparaginase schedule. Asparaginase-associated thrombosis is conventionally attributed to antithrombin III depletion from impaired hepatic protein synthesis - a mechanism with nothing to do with transsulfuration - and any CBS contribution would be additional to that and is entirely unevidenced.
- The actionable item is small, cheap and non-interventional: plasma homocysteine is a routine assay, and adding it to an existing correlative sample schedule would test the hypothesis without changing anyone's treatment. That is the only recommendation this entry supports.
- This entry concerns children and infants. Any onward use of it must go through the PI; a coordinator-generated mechanistic inference from a cell-line transcript is not a basis for contacting investigators or altering protocols.
- What is the prior probability that menin inhibition raises plasma homocysteine in a patient? CBS transcript falling 4.07 log2 in a leukaemic line is several inferential steps from systemic transsulfuration insufficiency, and without that prior the measurement gap is not obviously a defect.
- Does TINI 2 schedule ziftomenib and PEG asparaginase in overlapping blocks? The registry record does not resolve it, and the safety question only arises on co-exposure.
- Has any hyperhomocysteinaemia, thrombosis-above-background or unexplained thrombotic event been reported on a menin inhibitor? A FAERS query would answer it and was not run here - this analysis covers the registry only.
- Is CBS protein or enzyme activity affected at all, or only transcript? Nothing in this graph measures CBS protein under a menin inhibitor, and transsulfuration flux has never been measured in this disease.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
Across 57 distinct trials retrieved for nine menin-inhibitor names, plasma homocysteine appears as a named outcome measure in none, and neither does cystathionine, any amino-acid endpoint, any metabolomic endpoint, serine, folate or methionine. QTc appears in two. The outcome text analysed totals 113,991 characters with only two of the 57 trials carrying an empty outcome blob, and positive control tokens fire as expected - adverse 45/57, dose 34/57, survival 30/57, response 28/57, safety 21/57, remission 18/57 - so the zeros are results rather than an extraction failure.
What this evidence cannot show
AN OUTCOME-MEASURE SCAN IS NOT A PROTOCOL READ. Registry outcome measures are terse and routine safety chemistry is frequently not itemised, so a trial collecting a metabolic panel that happens to include homocysteine would not necessarily say so. The defensible claim is that homocysteine is not a NAMED endpoint in any of these 57 records - not that no sample has ever been assayed for it. `query.intr` is a text search that returns records merely mentioning a drug, so 57 is an upper bound on the trial set and some members are not menin-inhibitor trials. No full protocol or statistical analysis plan was retrieved for any trial.
Evidence that argues against this (2)
- The absence may be entirely appropriate rather than an oversight. Homocysteine is not a recognised toxicity endpoint for this drug class, no clinical signal of hyperhomocysteinaemia has been reported on a menin inhibitor, and adding an unvalidated analyte to a phase 1 dose-escalation protocol costs sample volume in children for a hypothesis generated from a cell-line transcript measurement. A trialist reading this would reasonably ask what the prior probability is, and this analysis cannot supply one.
- The QTc comparator cuts both ways. QTc is itemised because a QT signal was OBSERVED clinically for this class - which means the asymmetry demonstrates that these protocols respond to observed signals rather than to mechanistic predictions, and a mechanistic prediction from a transcript is exactly what this is. On that reading the absence of a homocysteine endpoint is the system working as designed, not failing.
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
The menin-inhibitor plus asparaginase combination is already being administered, including to infants. Three of the 57 trials mention asparaginase and all three list an asparaginase preparation among their interventions alongside a menin inhibitor. NCT05848687, 'TINI 2: Total Therapy for Infants With Acute Lymphoblastic Leukemia II', is RECRUITING, restricted to the CHILD age group, and lists PEG asparaginase and Ziftomenib among its interventions. NCT05761171 gives revumenib with chemotherapy including Calaspargase Pegol at a minimum age of 1 month in relapsed or refractory ALL and AML. NCT07636564 tests blinatumomab with or without revumenib alongside Calaspargase Pegol in B-cell ALL. Of the 57 trials, 17 mention lymphoblastic leukaemia, 44 mention KMT2A or MLL-rearranged disease, and 25 include the CHILD age group.
What this evidence cannot show
An intervention list establishes that both agents are given within the same protocol, NOT that they are given concurrently or in the same treatment block. TINI 2 is a multi-phase total-therapy backbone - induction, consolidation, reinduction blocks, maintenance - and ziftomenib may well occupy a different block from PEG asparaginase; the registry record does not resolve the schedule, and this entry does not claim temporal overlap. Two of the three trials are not yet recruiting or no longer recruiting. Nothing here establishes that any patient has received both agents.
Evidence that argues against this (2)
- This weakens the novelty of this graph's own hypothesis more than it strengthens the safety point. `hyp-menin-inhibition-collapses-the-serine-asns-programme-and-should-resensitise-kmt2a-r-all-to-asparaginase` proposes a combination that a recruiting infant trial is already delivering - so the therapeutic idea is not ours to propose, and the entry should be re-scoped to the mechanism rather than the combination.
- And the hypothesis it supports has been substantially undercut independently: Q109 measured ASNS's compound-selectivity margin at -0.25 with 86% of the effect reproduced by a DOT1L inhibitor, which puts ASNS on the compound-INDEPENDENT side. So the asparaginase-resensitisation prediction inherits none of the menin-selectivity it was built on, and a trial combining the two agents would be testing something this graph can no longer claim to have predicted.
CBS falls 4.07 log2 under menin inhibition, the largest fall in the serine panel, and germline CBS deficiency causes homocystinuria with thromboembolism - the gap this entry measures is the assay that would detect it
the combination this hypothesis proposes is ALREADY recruiting in infants (NCT05848687, PEG asparaginase + ziftomenib), so the therapeutic idea is not ours to propose - and the prediction independently lost its menin-selectivity when ASNS came back 86% reproduced by a DOT1L inhibitor
thrombosis is among its dose-limiting toxicities, conventionally via antithrombin III depletion; any CBS contribution would be additional to that and is unevidenced
co-listed with PEG asparaginase in NCT05848687, TINI 2, recruiting, infants with ALL - the schedule overlap is NOT resolved by the registry record and is not claimed
co-listed with Calaspargase Pegol in two trials, one from a minimum age of 1 month; and QTc - the recognised liability for this class - IS an itemised endpoint where homocysteine is not
acts on that beat's own most actionable item, which it flagged as a routine assay nobody has run