Blocking IKK/NF-kB in an infant is removing the switch every immune cell runs on — and the class has already failed clinically on exactly that
also called IKK inhibitor toxicity · NF-kB essential-function risk · the infant immune argument
NF-kB is the switch that immune cells use to respond to infection, and it is also what normal blood stem cells rely on under stress. Drugs that block it have repeatedly failed in people because of the side effects, and the patients here are babies whose immune systems are still being built while chemotherapy is already suppressing them.
The same thing, for a scientist
**This entry exists because everything else in this folder points at an intervention and the intervention has a documented class failure.** The statement is not a generic caution: a 2026 paper developing an alternative to IKKb inhibition says plainly that **pharmacological inhibition of IKKb 'has been limited by serious toxicities, including neutrophilia'**, and its preprint version says that **'nearly all IKKb inhibitors have failed prior clinical trials due to their serious side effects such as neutrophilia and systematic inflammation'** [E1]. That paper's entire contribution is a workaround — a cereblon PROTAC that degrades IKKb and NR4A1 and suppresses **KMT2A::MLLT3** AML in mice **without inducing neutrophilia** [E1]. So the toxicity is real enough that the field's response has been to change modality. **Four separate risk layers, in the order they would bite.** (1) **Infection.** Canonical NF-kB is the transcriptional output of every innate immune receptor; infection is the commonest cause of treatment-related death in infant ALL. (2) **Granulopoiesis.** The reported class effect is neutrophilia and systemic inflammation — a dysregulation of the myeloid compartment, not simply suppression of it. (3) **Normal stem cells.** The claimed window rests on constitutive NF-kB activation being a feature of primitive AML cells and not of normal primitive haematopoietic cells [E2], which is an observation in AML and not a measured therapeutic index in KMT2A-rearranged disease. (4) **Development.** Germline IKBKG loss-of-function in humans causes anhidrotic ectodermal dysplasia with immunodeficiency, and the non-canonical arm builds secondary lymphoid organs — these are developing systems in the index patient. **And the mirror risk.** The alternative strategy in this folder, NIK stabilisation, is not safer: it impairs normal HSPC self-renewal, which the authors name as the central obstacle, and it activates an arm that is oncogenic in B-cell malignancies [E3].
What the published selectivity evidence does and does not support
The founding paper reports differential effects between MLL leukaemia cells and normal haematopoietic progenitors across several assays: IKK inhibitors and Ikk knockdowns reduced growth and clonogenicity of MLL-transformed and MLL AML cells but not of normal progenitors or E2A-PBX1-transduced cells 2. That is real and it is in vitro colony biology in mice.
It does not establish a therapeutic index. It does not measure infection risk, mature immune function, thymic output, B-cell development, or any in vivo toxicity of sustained IKK inhibition in a growing animal. Reading a colony-assay sparing as a safety margin is the specific error this entry exists to block. The brief's rule about database facts applies here to a phenotype: a selective effect in a dish is a property of the assay, not a property of the patient.
The modality question, which is the only honest route forward
Three modalities are on the table and they carry different risk profiles.
- Small-molecule IKK inhibitor. Class failure documented 1. No clinical agent.
- Targeted degrader. The one agent with KMT2A in vivo data avoided neutrophilia in mice 1 — but it is built on celastrol, a promiscuous electrophile, and it degrades a second protein whose direction of effect in AML is itself disputed. A clean, selective IKKb degrader does not exist.
- Canonical-selective disruption at NEMO. Structurally the cleanest, because it spares the protective non-canonical arm; the only in vivo agent is a peptide tested in dogs (
gene-ikbkg). Delivery, not biology, is the constraint.
Any proposal in this graph that names an IKK-directed therapy should name its modality, because the risk statement changes completely between them.
What would have to be measured before this is a proposal rather than a hazard
In order: (i) does IKK inhibition evict the fusion in the index lineage at all (exp-ikk-plus-menin-inhibitor-occupancy-combination); (ii) at what fraction of the eviction dose does normal B-lymphopoiesis and myelopoiesis fail in a humanised or immunocompetent model; (iii) does combination with a menin inhibitor allow both to be given sub-maximally; (iv) what does sustained partial IKK blockade do to infection challenge in a juvenile animal. Nothing past (i) exists in the literature.
| kind | essential-function |
|---|---|
| severity | high - documented class failure in prior clinical trials, in a population already at high infection risk |
| mitigation | Targeted degradation rather than inhibition (one KMT2A::MLLT3 mouse precedent without neutrophilia); canonical-selective disruption at NEMO to spare the protective non-canonical arm; sub-maximal dosing in combination with a menin inhibitor, which requires the additivity in hyp-ikk-inhibition-is-a-menin-independent-chromatin-eviction-route to be true. |
| mirror risk | NIK stabilisation, the opposite intervention, impairs normal HSPC self-renewal and activates an arm oncogenic in B-cell malignancies |
affects
[ "gene-ikbkb", "gene-chuk", "gene-ikbkg", "protein-ikk-complex", "gene-map3k14" ]
reported class toxicities
[ "neutrophilia", "systemic inflammation" ]
population specific factors
[ "infants with a developing adaptive immune system", "concurrent myelosuppressive induction chemotherapy", "infection is a leading cause of treatment-related mortality in infant ALL", "germline IKBKG loss-of-function in humans causes immunodeficiency with ectodermal dysplasia" ]
- The class has already failed. An entry proposing IKK inhibition in this disease has to explain what is different this time, and the only published answer is 'use a degrader instead', on the strength of one mouse experiment.
- The selectivity evidence is colony assays in mouse cells. No animal in this literature was treated with an IKK inhibitor and assessed for toxicity.
- Both directions in this pathway carry an essential-function cost: blocking canonical NF-kB removes immune competence, and activating non-canonical NF-kB damages normal stem cells and mimics a B-cell oncogenic lesion.
- The mitigation that would make this tolerable - sub-maximal dosing in combination - depends on an additivity claim that has never been tested in any system.
- Infection is already the commonest cause of treatment-related death in this disease. A drug whose mechanism is removing the transcriptional response to infection is being proposed for the patients least able to absorb it.
- What fraction of the fusion-eviction dose can be given before normal B-lymphopoiesis fails? Nobody has measured the two on the same axis in any model.
- Does a degrader's kinetics - transient depletion with recovery between doses - open a window that continuous inhibition does not?
- Is the LSC-versus-HSC NF-kB dependence difference real in KMT2A-rearranged disease specifically, or is it an AML-wide observation being read across?
- What would sustained partial IKK blockade do to vaccine responses and to infection challenge in a juvenile animal? This is the experiment that would actually decide feasibility and nobody has run it.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Pharmacological inhibition of IKKb has been limited by serious toxicities including neutrophilia; the preprint version of the same work states that nearly all IKKb inhibitors have failed prior clinical trials due to serious side effects such as neutrophilia and systemic inflammation. The authors' response was to develop a celastrol-based cereblon PROTAC, compound A9, which degrades IKKb and NR4A1, suppresses leukaemia progression in a KMT2A::MLLT3 AML mouse model, and does not induce neutrophilia in vivo.
PROTAC-mediated targeting of IKKβ and NR4A1 for AML therapy. · Maharjan CK, Liu Y, ... Zhang W · Oncogene 2026What this evidence cannot show
Abstract read; full text not retrieved, and the statement about prior clinical failures is a framing sentence with its primary sources not enumerated in the abstract. It is cited here as the field's own published characterisation of the class, not as a systematic review of IKK trials. The 'without neutrophilia' claim is from a mouse model over the duration of that experiment, not a paediatric safety finding. The preprint is PMID 41040298, doi 10.1101/2025.09.24.678355.
Evidence that argues against this (1)
- That nearly all IKKb inhibitors failed clinical trials on toxicity is drawn in part from a PREPRINT version of the same work, so the strongest form of this safety statement is not peer-reviewed. Neutrophilia is also a mechanistically expected on-target effect rather than an idiosyncratic one, which makes it predictable but not avoidable by better chemistry. Habineza Ndikuyeze G et al. 2014
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
IKK inhibitors and shRNA knockdown of Ikka, Ikkb or Ikkg reduced growth and clonogenicity in MLL-oncogene-transduced and MLL AML cells but not in E2A-PBX1-transduced cells or normal haematopoietic progenitors; differential effects on colony formation and cell growth were observed in MLL leukaemia cells compared with normal haematopoietic progenitors. The authors note that constitutive NF-kB activation has been observed in primitive AML cells but not in normal primitive haematopoietic cells, and that indirect NF-kB inhibition by proteasome blockade induces LSC apoptosis while leaving normal HSCs viable, suggesting NF-kB dependence may distinguish LSCs from HSCs.
Epigenetic roles of MLL oncoproteins are dependent on NF-κB. · Kuo HP, Wang Z, Lee DF, Iwasaki M, Duque-Afonso J, Wong SHK, Lin CH, Figueroa ME, Su J, Lemischka IR, Cleary ML · Cancer Cell 2013What this evidence cannot show
All selectivity evidence is in vitro colony and growth biology in mouse cells; no animal was treated with an IKK inhibitor in this paper and no toxicity endpoint of any kind was measured. The LSC-versus-HSC distinction is cited by the authors from earlier AML literature, not demonstrated here. The proteasome-blockade argument is particularly weak support for an IKK proposal, since proteasome inhibition in MLL pro-B leukaemia works through stabilising MLL-AF4 rather than through NF-kB. Full text read this session (PMC3816582).
Evidence that argues against this (1)
- Systemic canonical NF-kB blockade has been achieved in a living animal without the feared haematological toxicity: intravenous NEMO-binding-domain peptide below 2 mg/kg inhibited constitutive canonical NF-kB in 6 of 10 dogs with ABC-like DLBCL, with only grade 1 toxicities in 3 and no significant haematological or biochemical toxicity. The class risk may be modality-specific rather than target-intrinsic. Habineza Ndikuyeze G et al. 2014
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The alternative strategy in this pathway carries its own essential-function risk: NIK stabilisation impairs the self-renewal of normal haematopoietic stem and progenitor cells, which the authors identify as the key challenge, proposing leukaemia-restricted delivery via LSC surface receptors such as TIM3 or CD93 as a currently unavailable solution.
Stabilization of NF-κB-Inducing Kinase Suppresses MLL-AF9-Induced Acute Myeloid Leukemia. · Xiu Y, Dong Q, Li Q, Li F, Borcherding N, Zhang W, Boyce B, Xue HH, Zhao C · Cell Reports 2018What this evidence cannot show
The HSPC impairment is cited by these authors to their own earlier work rather than shown afresh here. Full text read this session (PMC5810947).
Evidence that argues against this (1)
- Impaired self-renewal of normal HSPCs is measured in mice, where haematopoietic reserve and transplant stress differ from an infant's, and the authors identify it as the dose-limiting concern rather than measuring a therapeutic index. It is a reason for caution, not a quantified risk. Maharjan CK et al. 2026
- E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
A NEMO-binding domain peptide delivered intravenously below 2 mg/kg was safe in dogs with spontaneous activated-B-cell-like DLBCL and inhibited constitutive canonical NF-kB in 6 of 10, with grade 1 toxicities in 3 dogs and transient subclinical hepatopathy in 1; no long-term toxicities were identified and no significant haematological or biochemical toxicities were seen.
A phase I clinical trial of systemically delivered NEMO binding domain peptide in dogs with spontaneous activated B-cell like diffuse large B-cell lymphoma. · Habineza Ndikuyeze G, Gaurnier-Hausser A, ... Mason NJ · PLoS One 2014What this evidence cannot show
Abstract read this session. n=10, single-arm, adult dogs, a different disease, a peptide rather than a small molecule, and short exposure. It shows the node can be engaged systemically without acute catastrophe in one species; it says nothing about sustained blockade in an infant. Cited to bound the risk claim, not to soften it.
Evidence that argues against this (1)
- Ten dogs with a different disease, with inhibition achieved in 6 of 10 below 2 mg/kg, is a safety signal in an outbred large animal and nothing more. Canine DLBCL depends on canonical NF-kB in a way KMT2A-rearranged infant leukaemia is not shown to, so tolerability there does not predict either safety or efficacy here. Xiu Y et al. 2018
class failure in prior clinical trials on neutrophilia and systemic inflammation
the therapeutic half of the hypothesis is constrained by a documented class toxicity
the mirror risk: NIK stabilisation impairs normal HSPC self-renewal and mimics a B-cell oncogenic lesion
infection is a leading cause of treatment-related mortality in the index population