Four different ways to lose menin, and the graph has been treating them as one
also called germline versus pharmacological menin loss · menin exposure classes · MEN1 evidence transfer error
There are at least four different ways a cell can end up with less menin working, from inheriting a broken gene for life to taking a pill for a few months. They are not the same exposure, and evidence from one of them is not evidence about another.
The same thing, for a scientist
The single most consequential framing correction this beat needs. The `endocrine/` entries move freely between MEN1 carrier penetrance curves, conditional knockout mice, and drug-label laboratory abnormalities, as if they described one exposure at different resolutions. They do not. At least four distinct exposure classes exist, differing in depth, completeness, duration, reversibility and which menin functions are lost - and the empirical outcomes differ accordingly, in both directions [E1][E2][E3][E4][E5]. Evidence transfer between them is legitimate only when the transfer is argued, and in this beat it has usually been assumed.
The four classes, with what each one actually produces
| Class | What is lost | Duration | Documented outcome |
|---|---|---|---|
| 1. Physiological down-regulation | ~50% of menin protein, all functions, cell's own decision | Days to weeks, ends with the stimulus | Adaptive beta-cell proliferation; blocking it causes gestational diabetes 1 |
| 2. Acute induced biallelic excision | All menin, all functions, complete in targeted cells | Weeks to months (experimental) | Reverses established hyperglycaemia in three diabetic models 2 |
| 3. Germline monoallelic loss + somatic second hit | One allele from conception, second lost stochastically | Decades | MEN1: 7% penetrance at 10, 52% at 20; parathyroid first 3 |
| 4. Pharmacological pocket occupancy | Competitive block of one binding surface, partial, reversible | Median 2.5 months in the registration population; FDA's PMR specifies >=3 years | PTH up 34%, phosphate up 51%, ALP up 33%; no endocrine tumour reported 4 |
Class 2 is deeper than class 4 and produces benefit at the durations studied. Class 3 is shallower per cell than class 2 and produces tumours, because it runs for decades. So neither depth nor duration alone predicts the sign; the product does.
The specific transfers this beat has made, and whether they survive
- "Men1 beta-cell ablation gives insulinoma at 6 months, therefore a menin inhibitor threatens the pancreas." This is class 2 to class 4, and it fails in the one place it has been tested: 12 weeks of MI-463 at 70 mg/kg/day gave increased beta-cell mass with improved glucose handling 5. The transfer is not invalid, but it has a contrary datum and must be stated with it.
- "MEN1 carriers get hyperparathyroidism, therefore raised PTH on revumenib is on-target." Class 3 to class 4. It now has a mechanism that does not require the transfer at all - menin is required for TGF-beta restraint of PTH secretion, which operates on a timescale of weeks [see
mechanism-menin-is-required-for-tgfbeta-restraint-of-pth-secretion]. The conclusion survives; the argument for it should change. - "93% of MEN1 carriers are disease-free at age 10, therefore months of drug should be safe." Class 3 to class 4 in the reassuring direction, and it is no better than the alarming one. Carrier penetrance measures a different exposure and cannot bound a drug effect either way.
- "Hyperplasia in the rat toxicology package is what tumour-suppressor loss produces." True, and also what adaptive compensation produces 15. Hyperplasia is not a diagnosis.
The rule this beat should adopt
State the exposure class of every piece of evidence, and never let a claim inherit confidence from a different class. A MEN1 penetrance curve is evidence about MEN1 carriers. A conditional knockout is evidence about complete deletion. A label laboratory table is evidence about the drug. Where a transfer is needed - and it usually is, because the drug data are thin - the transfer is a claim in its own right, needs its own confidence, and should be flagged as the weakest link in whatever argument it supports.
classes
[
{
"name": "physiological down-regulation",
"depth": "~50% protein",
"duration": "days-weeks",
"sign": "adaptive"
},
{
"name": "acute induced biallelic excision",
"depth": "complete in targeted cells",
"duration": "weeks-months",
"sign": "beneficial early, neoplastic late"
},
{
"name": "germline monoallelic + second hit",
"depth": "one allele lifelong",
"duration": "decades",
"sign": "neoplastic"
},
{
"name": "pharmacological pocket occupancy",
"depth": "partial, one binding surface",
"duration": "median 2.5 months, PMR specifies >=3 years",
"sign": "laboratory abnormalities; no tumour reported"
}
]- The most reassuring statements in this beat and the most alarming ones both rest on the same unlicensed transfer between exposure classes, so tightening the rule removes comfort as well as alarm.
- The exposure distribution is moving from class 4 towards class 2 as menin inhibitors enter maintenance and front-line settings, which is the direction in which the transfers become more legitimate and the risk more real.
- Nobody has established that blocking a pocket is the same perturbation as removing the protein, and every argument in this beat assumes it in one direction or the other.
- Does pocket occupancy phenocopy reduced menin abundance transcriptionally? A side-by-side RNA-seq of menin knockdown versus a clinical menin inhibitor in the same endocrine cell would ground every transfer in this beat and has never been done.
- Is there a fifth class - intermittent or cycled dosing - that decouples the leukaemic effect from the endocrine one?
- At what cumulative exposure does class 4 start behaving like class 2?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Physiological menin down-regulation (~50%, pregnancy) drives adaptive beta-cell proliferation, and preventing it causes gestational diabetes in mice.
Menin controls growth of pancreatic beta-cells in pregnant mice and promotes gestational diabetes mellitus. · Karnik SK, Chen H, McLean GW, Heit JJ, Gu X, Zhang AY, et al. · Science 2007What this evidence cannot show
Mouse pregnancy; the magnitude figure is from a companion study.
Evidence that argues against this (1)
- Physiological down-regulation is bounded by a stimulus that ends, which is the one property pharmacological exposure does not share. Syndax Pharmaceuticals et al. 2026
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Acute inducible biallelic Men1 excision reverses established hyperglycaemia in three diabetic mouse models.
Reversal of preexisting hyperglycemia in diabetic mice by acute deletion of the Men1 gene. · Yang Y, Gurung B, Wu T, Wang H, Stoffers DA, Hua X · Proceedings of the National Academy of Sciences USA 2010What this evidence cannot show
Mouse; no long-term tumour follow-up.
Evidence that argues against this (1)
- The same class of exposure produces insulinoma and glucagon-expressing tumours on longer follow-up. Li F et al. 2015
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Germline monoallelic MEN1 loss produces parathyroid, pancreatic and pituitary tumours over decades, with hypercalcaemia and hypophosphataemia in the mouse counterpart.
Multiple endocrine neoplasia type 1 knockout mice develop parathyroid, pancreatic, pituitary and adrenal tumours with hypercalcaemia, hypophosphataemia and hypercorticosteronaemia. · Harding B, Lemos MC, Reed AA, Walls GV, Jeyabalan J, Bowl MR, et al. · Endocrine-Related Cancer 2009What this evidence cannot show
Mouse heterozygous model standing in for the human penetrance curve, which is cited separately in epi-men1-age-penetrance-curve.
Evidence that argues against this (1)
- The mouse biochemical phenotype (phosphate down) is opposite in direction to the human drug phenotype (phosphate up in 51%). Syndax Pharmaceuticals et al. 2026
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
On revumenib, intact PTH rose in 34%, phosphate in 51% and ALP in 33%, with a median exposure of 2.5 months and no endocrine tumour reported.
REVUFORJ (revumenib) tablets, for oral use - US Prescribing Information (SPL version 5, published 20 Feb 2026) · Syndax Pharmaceuticals, Inc. · DailyMed / FDA Structured Product Label, setid 6eb3cdbc-0e74-477d-82d6-3bb172d3f63f 2026What this evidence cannot show
Label laboratory table; median exposure is short and the distribution is lengthening as the drug moves into maintenance settings.
Evidence that argues against this (1)
- FDA's postmarketing requirement 4730-1 specifies patients treated for at least 3 years, so the exposure class the label describes is not the exposure class the regulator is worried about. Syndax Pharmaceuticals et al. 2026
- E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Twelve weeks of a menin inhibitor at 70 mg/kg/day increased beta-cell mass and improved glucose handling in diabetic mice, with no tumour assessment performed.
Menin-regulated Pbk controls high fat diet-induced compensatory beta cell proliferation. · Ma J, Xing B, Cao Y, He X, et al. · EMBO Molecular Medicine 2021What this evidence cannot show
Tool compound, n=4-5, adult mice, no histopathology for neoplasia and no off-drug follow-up.
Evidence that argues against this (1)
- Absence of a reported tumour in a study that never examined for one is absence of measurement, not evidence of safety. Syndax Pharmaceuticals et al. 2026
The latency entry brackets the question with two exposure classes (complete ablation, germline heterozygosity) and treats the drug as sitting 'in neither bracket'; there are at least four classes and the physiological and pharmacological ones both have data the entry omits
Makes the exposure-class rule explicit and gives it a table; that entry already frames the crux correctly and this one supplies the taxonomy
Class 3: the penetrance curve is evidence about germline carriers and cannot bound a drug effect in either the reassuring or the alarming direction
- ← depends onThe same drug target, pursued for benefit: menin inhibition is a diabetes strategy in the literature
The two literatures are reconciled only by exposure class and tissue, not by one of them being wrong