Eltanexor (KPT-8602) - the second-generation SINE whose whole point is that you can give it more often
also called KPT-8602 · ATG-016 · second-generation SINE
Eltanexor blocks the same nuclear export protein as selinexor but crosses much less into the brain, so it can be given far more frequently. That turns out to matter: in leukaemia models the difference between a weekly dose and a near-daily one is the difference between a temporary effect and a permanent one.
The same thing, for a scientist
Eltanexor is a second-generation selective inhibitor of nuclear export (SINE), covalent at **XPO1 Cys528** like selinexor but with markedly reduced central-nervous-system penetration, which permits **more frequent dosing**. Its relevance to this beat is not potency - it is **schedule**. **The schedule result.** Selinexor once or twice weekly produced only a **temporary** disruption of the XPO1-NPM1c interaction and **no clinical benefit** in NPM1-mutated patients. Eltanexor, dosed more frequently, produced **irreversible HOX down-regulation, terminal AML differentiation and prolonged survival** in NPM1-mutated models [E1]. That reframes every intermittent-schedule XPO1 result in leukaemia as a test of a different intervention. **Where it has been tested.** Registered trials found on ClinicalTrials.gov on 2026-09-20: **NCT02649790** (phase 1/2, completed, n=277, across relapsed/refractory myeloma, mCRC, mCRPC, higher-risk MDS, AML and newly diagnosed intermediate/high-risk MDS); **NCT06399640** (phase 1, recruiting, n=60, eltanexor plus venetoclax in relapsed/refractory MDS and AML); and **NCT05918055** (phase 1/2, **TERMINATED after 3 enrolled**, eltanexor with decitabine-cedazuridine in high-risk MDS, `whyStopped`: *'Drug company withdrew support.'*) [E2]. **In this beat's genotype.** t(6;9)/**DEK::NUP214** primary AML samples show a **selective response to selinexor and eltanexor** in a >500-compound screen across 57 primary AMLs [E3] - the strongest primary-sample XPO1 signal anywhere in the NUP214 literature. **Nothing has been tested in KMT2A-rearranged disease** with eltanexor, and there is **no paediatric eltanexor trial of any kind**. **Does it warrant a separate entry from `drug-selinexor`? Yes**, and specifically because the CNS-penetration difference is the axis on which a paediatric argument would turn: selinexor's paediatric dose ceiling was set by **reversible cerebellar toxicity**, and eltanexor is the compound engineered to avoid exactly that.
The paediatric argument that has not been made, and its missing piece
The chain is short and each link is documented:
- Selinexor is tolerable in children to 55 mg/m2; reversible cerebellar toxicity at 70 mg/m2 was dose-limiting (
drug-selinexor, NCT02212561). - Eltanexor was designed for reduced CNS penetration and is dosed far more frequently than selinexor.
- In NPM1-mutated AML, the frequency and not the potency is what converts a temporary XPO1-NPM1c disruption into irreversible HOX loss 1.
The missing piece is that there is no paediatric eltanexor data at all - no phase 1, no pharmacokinetics, no CNS exposure measurement in a child, and no infant anything. The three registered trials are adult. So the argument is coherent and entirely untested, and anyone making it should say which of the three links is inferred rather than measured: link 2 is a compound property, link 3 is a mouse result, and the paediatric extension is pure inference.
The terminated trial, and why the reason matters
NCT05918055 (eltanexor plus decitabine-cedazuridine in high-risk MDS) was terminated after 3 patients with whyStopped = 'Drug company withdrew support.' 2.
That string is worth recording precisely because it is not a safety or efficacy signal. A reader seeing 'terminated phase 1/2 in MDS' without the reason would reasonably infer toxicity. The registry says commercial withdrawal. This graph has twice found a whyStopped field overturning an entry's conclusion, and this is a third instance of the same class.
| target | gene-xpo1 |
|---|---|
| modality | oral small molecule, covalent SINE at XPO1 Cys528 |
| status | investigational; no regulatory approval found as of 2026-09-20; no paediatric trial of any kind |
| distinguishing property | markedly reduced CNS penetration relative to selinexor, permitting more frequent dosing |
| kmt2a specific evidence | none - no preclinical or clinical eltanexor study in a KMT2A-rearranged system was found on 2026-09-20 |
| pediatric evidence | none |
trials
[ "NCT02649790", "NCT06399640", "NCT05918055" ]
trial notes
{
"NCT02649790": "phase 1/2, completed, n=277, multi-tumour including AML and higher-risk MDS",
"NCT06399640": "phase 1, recruiting, n=60, eltanexor + venetoclax, relapsed/refractory MDS and AML",
"NCT05918055": "phase 1/2, TERMINATED after 3 enrolled, eltanexor + decitabine-cedazuridine, high-risk MDS; whyStopped = 'Drug company withdrew support.'"
}genotypes with reported preferential activity
[ "NPM1-mutated AML", "DEK::NUP214 / t(6;9) AML", "SF3B1-mutated myeloid neoplasms" ]
key ae
[ "thrombocytopenia and neutropenia (SINE class)", "nausea, anorexia, weight loss (SINE class)", "hyponatraemia (SINE class)" ]
resistance
[ "not characterised in any HOX-driven leukaemia; XPO1 C528S is the canonical laboratory escape mutation for covalent SINEs and has not been reported clinically in the sources read here" ]
- More frequent dosing of a common-essential target's inhibitor means more cumulative exposure of normal tissue, and the SINE class toxicities - anorexia, weight loss, thrombocytopenia - are cumulative and schedule-dependent.
- There is no paediatric safety, pharmacokinetic or CNS-exposure data for eltanexor. The whole reason to prefer it over selinexor in a child is a compound property measured in adults and animals.
- One of its three registered trials stopped because the sponsor withdrew support. Commercial availability is a real risk for an investigational agent with no approval, independent of the biology.
- The 2026 field review lists the genotypes where XPO1 inhibitors work and KMT2A rearrangement is not among them. Proposing it here means arguing against the field's own current summary.
- Is eltanexor active in a KMT2A-rearranged model? The compound exists, the assay is a one-week viability experiment, and nobody has published it.
- What is eltanexor's CNS exposure in a child? The entire paediatric argument rests on it and there is no measurement.
- If frequency rather than potency is what matters, does a continuous low-dose selinexor schedule reproduce the eltanexor effect? That would be cheaper to test than a new paediatric compound.
- Why does t(6;9) respond selectively while other NUP214 fusions have not been screened the same way? SET::NUP214 has never appeared in a primary-sample XPO1-inhibitor drug screen.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Selinexor administered once or twice per week results in only a temporary disruption of the XPO1-NPM1c interaction and did not translate into clinical benefit in NPM1-mutated patients; because eltanexor can be administered more frequently, prolonged XPO1 inhibition with eltanexor caused irreversible HOX down-regulation, induced terminal AML differentiation and prolonged survival of leukaemic mice in NPM1-mutated AML models.
Prolonged XPO1 inhibition is essential for optimal antileukemic activity in NPM1-mutated AML. · Pianigiani G, Gagliardi A, Mezzasoma F, et al. · Blood Advances 2022What this evidence cannot show
Abstract-level for this beat. NPM1-mutated AML models, adult myeloid disease; no lymphoid, paediatric or KMT2A-rearranged data. 'Irreversible' is the authors' word and rests on preclinical models. The clinical half of the claim summarises prior selinexor experience rather than reporting a trial.
Evidence that argues against this (1)
- The one randomised test of the first-generation compound in acute leukaemia was negative and numerically unfavourable, median OS 3.2 versus 5.6 months, HR 1.18. The schedule argument explains that result away, and an explanation constructed after a negative trial is a hypothesis about the trial, not a second trial. No randomised eltanexor data in leukaemia exist. Sweet K et al. 2021
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.trial_registry
ClinicalTrials.gov lists three eltanexor studies as of 2026-09-20: NCT02649790 (phase 1/2, completed, enrolment 277, in relapsed/refractory multiple myeloma, metastatic colorectal cancer, metastatic castration-resistant prostate cancer, higher-risk MDS, AML and newly diagnosed intermediate/high-risk MDS); NCT06399640 (phase 1, recruiting, enrolment 60, eltanexor and venetoclax in relapsed or refractory MDS and AML); and NCT05918055 (phase 1/2, terminated after 3 enrolled, eltanexor with decitabine-cedazuridine in high-risk MDS, whyStopped 'Drug company withdrew support.').
ClinicalTrials.gov API v2 query for intervention 'eltanexor' · ClinicalTrials.gov · ClinicalTrials.gov 2026What this evidence cannot show
Registry snapshot taken 2026-09-20 via the raw API (not the MCP tool, which does not return whyStopped). Registry fields are sponsor-supplied and can lag. Enrolment counts are the registry's `enrollmentInfo.count`, which may be actual or anticipated depending on status. NONE of the three trials is paediatric and none selects on KMT2A rearrangement.
Evidence that argues against this (1)
- PUBLISHED CONFLICTING RESULT against the schedule premise that makes this registry snapshot matter. This entry's framing is that eltanexor exists because intermittent selinexor dosing is insufficient. But selinexor given TWICE WEEKLY with daunorubicin and cytarabine produced CR/CRi in 8 of 19 evaluable previously untreated poor-risk AML patients, with the MTD not reached and median overall survival 10.3 months. An intermittent schedule is therefore not inert in AML; what the NPM1c work establishes is that intermittent dosing does not sustain disruption of one particular protein-protein interaction, which is a narrower claim than 'schedule is the whole point'. Sweet K et al. 2020
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In a multi-omics study of 57 primary AML samples including drug screening with more than 500 compounds, t(6;9)/DEK::NUP214 cases displayed a selective response to the XPO1 inhibitors selinexor and eltanexor, together with a transcriptomic signature of FOXC1 and HOX overexpression.
DEK::NUP214 acts as an XPO1-dependent transcriptional activator of essential leukemia genes. · Kaya F, Bewicke-Copley F, Miettinen JJ, et al. · Leukemia 2025What this evidence cannot show
Abstract-level for this beat. PRIMARY PATIENT SAMPLES, which is rare in this literature, but t(6;9) is a rare entity so the number of index cases within the 57 is not stated in the abstract and could be very small. Ex vivo drug screening on primary AML is a viability readout and does not establish a clinical response rate. Adult/younger-adult AML, not paediatric and not KMT2A-rearranged. SCOPE CORRECTED 2026-09-21. The abstract of PMID 40204893 was re-read in full. The 57 primary AML samples are the MULTI-OMICS AND DRUG-SCREEN arm only - whole genome sequencing, targeted sequencing, transcriptomics and >500-compound screening - and that arm is what shows the selective t(6;9) response to selinexor and eltanexor. The CUT&RUN is a SEPARATE experiment with no n stated in the abstract, and the key result that DEK::NUP214 binding falls under XPO1 inhibition is stated as occurring 'in t(6;9) cells', not in the 57 primary samples. Attributing the occupancy readout to 57 primary patient samples makes it a much stronger claim than the paper supports. The conflation entered via a hand-back summary and was then repeated in a dispatch brief before being caught by the beat that received it.
Evidence that argues against this (1)
- NAMED METHODOLOGICAL DEFEATER: the specificity control for attributing anything to XPO1 inhibition is the drug-resistant allele, and an ex vivo viability screen has none. Chen et al. make this the explicit standard - every C646 phenotype they attribute to XPO1 is shown to be abrogated in cells carrying the XPO1-C528S drug-resistance allele, and they highlight the utility of resistant alleles for exactly this purpose. A selective viability response of t(6;9) cases in a >500-compound screen identifies a correlation between a genotype and two compounds; without a C528S arm it does not establish that the killing runs through XPO1 rather than through an off-target liability those two compounds share. Chen YF et al. 2024
- E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.review
A 2026 review of XPO1 targeting in AML reports that selinexor and eltanexor show preferential activity in NPM1-mutated, DEK::NUP214-positive and SF3B1-mutated myeloid neoplasms, that combinations with hypomethylating agents and BCL-2 inhibitors enhance depth and durability of responses but are limited by toxicity, and that optimising dosing schedules and identifying predictive biomarkers remain the open problems.
Targeting the nuclear export receptor exportin-1 in acute myeloid leukaemia: From biology to clinical translation. · Liu Y, Yun X, Ding W, et al. · Clinical and Translational Medicine 2026What this evidence cannot show
A REVIEW, read at abstract level, cited as a map of the field's own summary of where XPO1 inhibitors work. Note what its list of responsive genotypes does NOT contain: KMT2A rearrangement. That absence in a 2026 review is the clearest statement available that this genotype is not currently considered an XPO1-inhibitor indication.
Evidence that argues against this (1)
- PUBLISHED CONFLICTING RESULT on why KMT2A rearrangement is missing from the review's list of responsive genotypes: nobody has looked in a way that could have put it there. The only paediatric selinexor leukaemia trial enrolled 18 children with relapsed or refractory acute leukaemia without selecting on genotype, and its endpoints were the MTD and recommended phase 2 dose of selinexor with fludarabine and cytarabine. A review can only list genotypes that someone enrolled and analysed. Reading the omission as 'the clearest statement available that this genotype is not an XPO1-inhibitor indication' converts an absence of testing into a negative result, which is the error this project's brief names explicitly. Alexander TB et al. 2016
prolonged XPO1 inhibition gives irreversible HOX down-regulation where weekly selinexor does not
selective ex vivo response of t(6;9) primary AML to selinexor and eltanexor
covalent second-generation SINE at Cys528, dosed more frequently
- analogous to →Selinexor (XPO1 inhibition) — a real paediatric response rate attached to an unproven mechanism in KMT2A-rearranged disease
same target and chemistry, different CNS penetration and schedule
- ← related toTwo reasons XPO1 is a hard target: every cell needs it, and blocking it switches on a survival pathway
the same class constraints, with no paediatric data at all
the premise that eltanexor exists because intermittent dosing is insufficient does not generalise: twice-weekly selinexor with daunorubicin and cytarabine gave CR/CRi in 8 of 19 evaluable previously untreated poor-risk AML patients, MTD not reached (PMID 31636097)