Measure ABCA1, ABCG1 and SREBF1 before and during menin inhibition - a three-transcript panel on samples that already exist
also called LXR PD panel menin inhibitor · revumenib lipid pharmacodynamics
Take blood samples already stored from patients before and after they started a menin inhibitor, and measure three genes that menin is supposed to keep switched off. If they go up, the drug has an effect on fat and cholesterol handling that nobody has recorded.
The same thing, for a scientist
A paired pre-dose / on-treatment measurement of the LXR target transcripts **ABCA1, ABCG1 and SREBF1** in banked PBMC or blast material from menin-inhibitor-treated patients, with a matched serum lipid panel where one exists, and a parallel in vitro arm in a hepatocyte line at clinically achievable exposures with an LXR antagonist as the specificity control. **The design is a within-patient paired comparison**, because the population's baseline lipid state is deranged by asparaginase and corticosteroids and a between-patient comparison cannot see through that [E2]. The readouts and their expected direction are already published in a non-haematopoietic system [E1].
Design
Arm A, clinical, paired. Pre-dose and on-treatment (cycle 1 day 15 or the nearest banked timepoint) PBMC or sorted blasts from the same patient. RT-qPCR or RNA-seq for ABCA1, ABCG1, SREBF1, with FASN and SCD-1 as secondary. Normalisers chosen against the graph's own warning that ACTB is a loading control and not a gene claim. Matched serum triglycerides and total cholesterol where the trial collected them.
Arm B, control for the backbone. The same paired panel in patients on the identical chemotherapy backbone WITHOUT a menin inhibitor. Without this arm the result is uninterpretable, because asparaginase and corticosteroids move the same analytes.
Arm C, in vitro specificity. HepG2 and a KMT2A-rearranged line (SEM or MV4-11), revumenib at clinically achievable plasma concentrations, +/- an LXR antagonist. The antagonist arm is what separates 'menin inhibition raises these transcripts' from 'menin inhibition raises these transcripts THROUGH LXR'.
Arm D, isoform attribution. NR1H2 and NR1H3 knockdown in the leukaemia line, to test prediction 5 - that the haematopoietic effect, if any, is LXR-beta-dependent.
Why the backbone arm is not optional
Every patient in these trials received cytotoxics, and the paediatric ALL backbone in particular includes asparaginase and dexamethasone, both of which raise triglycerides substantially on their own. A single-arm on-treatment rise would be attributed to the backbone by any reviewer, correctly. This is the same reasoning the graph applies to every unstated-system problem: a measurement without its comparator is a description of the population, not of the drug.
What each outcome would mean
| outcome | reading |
|---|---|
| Transcripts rise, backbone arm flat, antagonist abolishes it | the hypothesis holds and menin inhibitors are partial LXR agonists in patients |
| Transcripts rise in both arms | it is the backbone; the hypothesis's clinical arm is dead and the cell arm still stands |
| Nothing moves in arm C at clinically achievable exposure | the hypothesis is falsified - see the falsifier list on the hypothesis entry |
| Arm C moves, arm A does not | the mechanism is real and the exposure or the tissue is wrong; the hepatic question stays open |
| design | Arm A: paired pre-dose vs on-treatment PBMC or sorted blasts from menin-inhibitor-treated patients, RT-qPCR/RNA-seq for ABCA1, ABCG1, SREBF1 (primary) and FASN, SCD-1 (secondary), plus matched serum triglycerides and total cholesterol. Arm B: identical paired panel in patients on the same chemotherapy backbone without a menin inhibitor. Arm C: HepG2 and a KMT2A-rearranged line (SEM or MV4-11) treated with revumenib at clinically achievable plasma concentrations, +/- an LXR antagonist. Arm D: NR1H2 and NR1H3 knockdown in the leukaemia line to attribute the isoform. |
|---|---|
| readout | paired within-patient fold change in ABCA1/ABCG1/SREBF1, with the backbone arm as the comparator; in vitro, fold change abolished or not by LXR antagonism |
| primary endpoint | within-patient log2 fold change in ABCA1 and ABCG1, menin-inhibitor arm versus backbone-only arm |
| why now | revumenib is approved including in children, samples are banked, and the transcripts are three qPCR assays |
| cost class | low - retrospective on existing material plus one cell-line experiment |
| tests hypothesis | hyp-menin-inhibition-derepresses-lxr-target-genes-in-patients |
| status | proposed, not run |
controls
[ "backbone-only patient arm", "LXR antagonist in vitro", "PPAR-alpha target genes, to detect cancellation rather than absence", "isoform knockdown for attribution" ]
confounders
[ "asparaginase and corticosteroids raise triglycerides independently", "disease burden itself alters lipid profiles", "PBMC composition changes on treatment, so a sorted-blast arm is preferable where cell numbers allow" ]
- A positive result would add a metabolic warning to a drug that is currently one of the few good options in relapsed KMT2A-rearranged leukaemia. The finding must be reported with its effect size and not as a signal, or it will be read as a reason to withhold an effective agent.
- A negative arm-A result with a positive arm-C result is the likeliest outcome and is the least useful one. The design should be powered on the in vitro arm first.
- This is a retrospective analysis of banked material with all the selection problems that implies: the patients with banked paired samples are not a random sample of the treated population.
- Which menin-inhibitor trials banked paired pre-dose and on-treatment cellular material rather than serum alone? The design is only cheap where paired cells exist.
- What plasma concentration of revumenib should arm C use? The clinical exposure is published and the colorectal paper's inhibitor concentration is not, so the two cannot currently be matched.
- Should the panel include the PPAR-alpha targets as a co-primary rather than a control? A cancellation result is as interesting as a positive one and the current design treats it as a nuisance.
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
The readouts and their expected direction are published: menin inhibition upregulated ABCG1 and ABCA1 and reduced cellular cholesterol in colorectal cancer lines and mouse colonic epithelium, and menin overexpression reduced SREBP-1c, FASN and SCD-1 in hepatocytes.
Menin Maintains Cholesterol Content in Colorectal Cancer via Repression of LXR-Mediated Transcription. · Nyul TE, et al. · Cancers (Basel) 2023What this evidence cannot show
Abstract only; colorectal and hepatic systems, not haematopoietic (brief rule 73). The menin inhibitor used and its concentration are not named in the abstract, so the arm-C exposure would have to be set from clinical pharmacokinetics rather than from this paper.
Evidence that argues against this (2)
- The published direction may be an off-target direction, and that changes what arm C can conclude. Phillips et al. 2019 identify LANOSTEROL SYNTHASE as the direct molecular target of MI-2 and show its glioma activity is largely menin-independent, acting by suppressing cholesterol synthesis and generating the endogenous LXR ligand 24,25-epoxycholesterol. Since the colorectal paper does not name its compound, the 'expected direction' this experiment is designed around could be the pharmacology of a sterol-pathway enzyme inhibitor rather than of menin loss. Two consequences for the design: arm C should use revumenib (a chemically unrelated series) rather than an MI-series tool compound, and the LXR-antagonist control does NOT discriminate the two mechanisms, because an off-target rise in an endogenous LXR ligand is also abolished by an LXR antagonist. The discriminating control is the genetic one - MEN1 knockdown versus drug. Phillips RE et al. 2019
- The expected direction is also not the only one on record. In intestinal epithelium, Men1 deletion and the menin inhibitor MI-463 de-repress Ces1g through LXR-BETA and HDAC1, increasing triglyceride hydrolysis and fatty-acid oxidation and PROTECTING mice against diet-induced obesity. A three-transcript ABCA1/ABCG1/SREBF1 panel would not see that axis at all, so a flat arm A is compatible with a real and large menin-inhibitor lipid effect running through a gene this panel does not measure. CES1 belongs in the panel. Cao X et al. 2026
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
In HepG2 and mouse liver, menin represses LXR-alpha and lowers SREBP-1c, FASN and SCD-1 - the hepatic half of the prediction the clinical panel would test.
Tumor suppressor Menin acts as a corepressor of LXRα to inhibit hepatic lipogenesis. · Cheng P, et al. · FEBS Lett 2015What this evidence cannot show
Abstract read this session; Elsevier full text unavailable (HTTP 403 on DOI resolution). Transfection-based; no haematopoietic system, no KMT2A fusion (brief rule 73).
Evidence that argues against this (1)
- Menin co-activates PPAR-alpha in the same tissue, so a null result in arm A is compatible with both mechanisms being real and cancelling. The experiment should therefore also record PPAR-alpha targets, or it cannot distinguish 'no effect' from 'two effects'. Cheng P et al. 2011
the designed test, with a backbone comparator arm
tests whether the hepatic/intestinal mechanism operates in patients
the agent whose banked samples make this cheap