Oncofusion

start / graph / drug-dexamethasone

Drugstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q51-backbone

Dexamethasone - the steroid an infant is actually on for most of the protocol, and the one with a high-certainty neurodevelopmental signal in babies

also called dexamethasone · DEXA · DXM · Decadron · dexamethasone sodium phosphate · 9-fluoro-16alpha-methylprednisolone

Dexamethasone is the steroid that takes over from prednisone on day 8 and runs for most of induction and again in the reinduction block. In premature babies given it for lung disease, randomised trials show it raises the risk of cerebral palsy - which is a different situation, but it is the only high-quality evidence about what this drug does to a developing infant brain.

The same thing, for a scientist

In Interfant-06 dexamethasone is **6 mg/m2/day** PO or IV on induction days 8-28 with a one-week taper, and again **6 mg/m2/day** on OCTADA(D) part I days 1-14 tapering to zero by day 21 [E1]. It therefore accounts for far more steroid-days than prednisone does, while carrying none of the protocol's prognostic weight. Pharmacologically it is a fluorinated glucocorticoid with no 11-keto prodrug step, higher receptor affinity and longer half-life than prednisolone, and greater CSF penetration - the property that made it standard in childhood ALL. **The KMT2A-rearranged evidence for dexamethasone is class-level, not agent-level:** the resistance literature in MLL-rearranged infant ALL names 'glucocorticoids, for example prednisone and dexamethasone' and then measures prednisolone [E2][E3]. The one agent-level measurement this beat found is that romidepsin synergised with dexamethasone in only **three of six** infant ALL cell lines, against six of six for daunorubicin and cyclophosphamide [E4]. **The infant-specific evidence is the strongest in this beat and it comes from outside oncology:** in 32 randomised trials of early systemic postnatal corticosteroids in preterm infants, dexamethasone increased cerebral palsy with a risk ratio of 1.77 (95% CI 1.21-2.58, high-certainty GRADE) while hydrocortisone did not [E5].

Go deeper
Where it sits in the backbone
phasedexamethasone
remission induction6 mg/m2/day PO or IV, days 8-28, then tapered to 0 over one week
OCTADA(D) part I6 mg/m2/day PO or IV, days 1-14, tapered to 0 by day 21

That is roughly five weeks of systemic dexamethasone against one week of prednisone 1. The body-surface-area dose is multiplied by 2/3 under 6 months and 3/4 from 6 to 12 months, and the protocol states explicitly that the reduction applies to the glucocorticoids 1.

A registry defect worth recording: the ClinicalTrials.gov intervention list for NCT00550992 names prednisone, prednisolone and hydrocortisone and does not list dexamethasone at all, although the same record's detailed description prescribes it on days 8-28 6. Anyone enumerating the drugs of this protocol from the structured intervention field will miss the steroid the infant spends most of her induction on. This is the same failure shape this project has logged elsewhere: a well-formed, correctly named, incomplete record.

What is and is not known about dexamethasone specifically in this disease

Class-level: strong. Every statement in the MLL-rearranged infant ALL resistance literature about 'glucocorticoid resistance' is framed to include dexamethasone by name 23, and the four mechanisms are set out in drug-prednisolone.

Agent-level: thin, and the one measurement is discouraging. Cheung et al. scored romidepsin synergy against each Interfant backbone agent across six infant ALL cell lines: synergy in 6/6 for daunorubicin and 4-hydroperoxycyclophosphamide, 4/6 for vincristine, 3/6 for dexamethasone, and antagonism in 4/6 for methotrexate and L-asparaginase 4. Dexamethasone sits in the middle and is the only steroid tested.

The prednisone-versus-dexamethasone question has never been asked in infants. In older children it is one of the most-studied questions in paediatric oncology. Interfant-06 simply uses both, sequentially, and no trial in this age group has compared them. This beat records that as a bounded absence in E7 rather than importing the answer from the 1-18 year literature, because the infant population differs in exactly the organ - the developing brain - where the two agents are reported to differ.

The neurodevelopmental signal, and exactly how far it can be carried

The Cochrane review of early (<7 days) systemic postnatal corticosteroids in preterm infants pools 32 randomised trials, 4,395 infants, with high-certainty GRADE on the major outcomes 5. Dexamethasone reduced bronchopulmonary dysplasia (RR 0.72, 0.63-0.82) and increased cerebral palsy (RR 1.77, 95% CI 1.21-2.58) and gastrointestinal perforation (RR 1.73, 1.20-2.51). Hydrocortisone, in the same review, did not increase cerebral palsy (RR 1.05, 0.66-1.66) and did reduce mortality.

Now the scope limits, stated as forcefully as the finding. These are preterm infants, many extremely preterm, treated in the first six days of life for lung disease, at doses and schedules from the BPD literature, not 6 mg/m2/day for five weeks for leukaemia. The comparator is no steroid. Nothing here measures dexamethasone in an infant with ALL, and a leukaemia protocol without a steroid is not an available option.

What it does license. Three things. First, that the developing infant brain is a documented target organ for this specific molecule, at high certainty, in a way that distinguishes it from hydrocortisone. Second, that the dexamethasone-versus-prednisolone choice in infants is not the neutral pharmacokinetic convenience it is usually treated as. Third, that long-term neurocognitive follow-up is the outcome this population's trials do not report and should: the Cochrane authors' own conclusion is that longer-term follow-up into late childhood is vital for higher-order neurological functions that cannot be assessed early. Interfant-06 reports EFS and OS.

Facts and figures
targetNR3C1 glucocorticoid receptor (direct ligand; fluorinated, no prodrug step). No gene node for NR3C1 exists in this graph as of cycle 28.
modalitysmall molecule, fluorinated synthetic glucocorticoid
interfant06 positionremission induction 6 mg/m2/day PO or IV days 8-28 with a 1-week taper; OCTADA(D) part I 6 mg/m2/day days 1-14 tapering to 0 by day 21
interfant06 dose6 mg/m2/day
infant dose adjustmentx 2/3 under 6 months, x 3/4 from 6 through 12 months; the protocol states explicitly that the reduction applies to glucocorticoids
steroid days vs prednisoneabout five weeks of dexamethasone against one week of prednisone
kmt2a specific evidenceCLASS-LEVEL ONLY. The MLL-rearranged infant ALL glucocorticoid-resistance literature names dexamethasone and measures prednisolone. The single agent-level datum found is romidepsin synergy in 3 of 6 infant ALL cell lines.
infant specific toxicityCerebral palsy RR 1.77 (95% CI 1.21-2.58, high-certainty GRADE) with early systemic dexamethasone in preterm infants, against RR 1.05 (0.66-1.66) for hydrocortisone, in 32 randomised trials of 4,395 infants. Different population, different indication, different schedule - see E5 caveats.
registry defectClinicalTrials.gov NCT00550992 omits dexamethasone from its structured intervention list while prescribing it in the detailed description
key ae
[
  "hyperglycaemia requiring insulin",
  "hypertension",
  "infection",
  "osteonecrosis",
  "myopathy",
  "behavioural disturbance",
  "adrenal suppression"
]
resistance
[
  "the four prednisolone-resistance mechanisms reported in MLL-rearranged infant ALL are asserted to apply (MCL-1, S100A8/A9, annexin A2/Src, PI3K) but none was measured with dexamethasone",
  "NR3C1 pathway lesions are a documented route to glucocorticoid resistance in relapsed ALL generally and were not assessed in the infant studies cited here"
]
trials
[
  "NCT00550992",
  "NCT02553460"
]
How this could bite us back
  • **The single best-quality evidence about dexamethasone in babies says it raises cerebral palsy risk, and the leukaemia field has no comparable outcome data at all.** An infant cured of KMT2A-rearranged ALL has sixty years of neurological function ahead of her, and the trials that cured her measured event-free survival.
  • Steroid resistance is the defining pharmacological feature of this disease and the obvious response - more steroid - is aimed at the organ with the documented developmental toxicity, in the age band where that toxicity was measured.
  • The hydrocortisone-versus-dexamethasone separation in the neonatal data is a warning against class-level reasoning, and this graph's KMT2A resistance literature is entirely class-level for dexamethasone. A resensitisation strategy validated on prednisolone may simply not transfer to the steroid the infant is actually on for five weeks.
  • Any automated enumeration of this protocol's drugs from the ClinicalTrials.gov intervention field silently omits dexamethasone. A graph that built its infant drug inventory that way would have a hole exactly where the largest steroid exposure sits - and the same field is correct about doxorubicin's absence, so the field's silence carries no signal.
  • Dexamethasone's advantage over prednisolone in older children is CNS penetration, which is also the property that delivers it to a brain that is still myelinating. The therapeutic rationale and the toxicity concern are the same pharmacological fact.
Still open
  1. Does dexamethasone at antileukaemic doses affect neurodevelopment in infants treated for ALL? Interfant-06 reports EFS and OS; no infant leukaemia trial in this graph reports a neurocognitive outcome, and the only high-certainty evidence about this molecule in this age group comes from neonatology.
  2. Is any of the four reported prednisolone-resistance mechanisms actually operative against dexamethasone? The protocol gives five weeks of dexamethasone and one week of prednisone, and the mechanism literature has measured the drug that is given for one week.
  3. Would hydrocortisone-based induction be antileukaemically inferior in infants? The neonatal data separate hydrocortisone from dexamethasone on exactly the outcome that matters over sixty years, and nobody has asked what it would cost in leukaemia control.
  4. Why is a 6 mg/m2/day dexamethasone dose used in Interfant and 10 mg/m2/day in the COG infant study? The difference is nearly two-fold in an age group where the dose is then multiplied by 2/3, and no comparison exists.
Every source behind this entry (7)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Interfant-06 gives dexamethasone 6 mg/m2/day PO or IV on induction days 8-28 with a taper to zero over one week and 6 mg/m2/day on OCTADA(D) part I days 1-14 with a taper to zero by day 21; the body-surface-area dose is multiplied by 2/3 for infants under 6 months and 3/4 for those 6 through 12 months, a reduction the protocol states applies to all drugs including the glucocorticoids; for hypertension not controlled by antihypertensives and sodium restriction the protocol permits reducing the glucocorticoid dose by 30-50%.

    What this evidence cannot show

    A trial protocol document, not a peer-reviewed publication, and specifically the SKION (Dutch) national version of April 2006; the trial was amended during its course (asparaginase provisions in MARMA and OCTADA(D) among others), so later versions may differ. Doses given are PLANNED doses; delivered cumulative dose per patient is lower wherever a course was omitted, delayed or curtailed by a stopping rule, and the trial report does not publish delivered dose per patient. Retrieved and text-extracted by this beat on 2026-09-20 (90 pages, 194,979 characters); page numbers are the PDF's own.

    Evidence that argues against this (1)
    • A different contemporary infant ALL protocol uses a substantially higher dexamethasone dose: the COG-affiliated phase 1/2 study of bortezomib and vorinostat gave dexamethasone 5 mg/m2 twice daily (10 mg/m2/day) on days 1-4, 8-11 and 15-18 of induction. So 6 mg/m2/day is Interfant's choice, not a standard, and cross-protocol comparisons of steroid exposure in infants are not safe. Gruber TA et al. 2026
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    MLL-rearranged infant ALL is characterised in the primary literature as resistant to glucocorticoids naming prednisone and dexamethasone together, while the experimental measurements reported in those same papers are made with prednisolone.

    Elevated S100A8/S100A9 expression causes glucocorticoid resistance in MLL-rearranged infant acute lymphoblastic leukemia. · Spijkers-Hagelstein JA, Schneider P, Hulleman E, de Boer J, Williams O, Pieters R, et al. · Leukemia 2012
    PMID 22282267DOI
    What this evidence cannot show

    This is a claim about how the literature is worded and what it measured, verified by reading the abstracts of this paper and of PMID 23334362 and 19965632. It is not a claim that dexamethasone resistance has been disproved in this disease - only that it has not been separately measured in the studies that name it.

    Evidence that argues against this (1)
    • Glucocorticoids are not interchangeable even within the class. In the same Cochrane data set, dexamethasone and hydrocortisone differ in direction on mortality and on cerebral palsy, so class-level reasoning across glucocorticoids has a documented failure mode in infants specifically. Doyle LW et al. 2021
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In MLL-rearranged infant ALL, high ANXA2 with Src-kinase phosphorylation of annexin A2 discriminates glucocorticoid-resistant from glucocorticoid-sensitive patient samples, and the paper frames the resistance as being to glucocorticoids 'like prednisone and dexamethasone' while measuring prednisolone.

    Src kinase-induced phosphorylation of annexin A2 mediates glucocorticoid resistance in MLL-rearranged infant acute lymphoblastic leukemia. · Spijkers-Hagelstein JA, Mimoso Pinhanços S, Schneider P, Pieters R, Stam RW · Leukemia 2013
    PMID 23334362DOI
    What this evidence cannot show

    In vitro only, single group, and the dexamethasone framing is in the introduction rather than in the results. Four independent knockdowns converging is good internal evidence for the mechanism; none of them was tested against dexamethasone.

    Evidence that argues against this (1)
    • An independent group working in MLL-ALL patient-derived xenografts reached a partly different conclusion - BCL-2 dependence revealed by BH3 profiling, with the kinase inhibitor selecting for it - and their in-vivo glucocorticoid is not resolvable from the abstract either. The ambiguity about which steroid was used runs through the whole literature, not just one laboratory. de Groot AP et al. 2021
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Across a panel of six infant ALL cell lines, romidepsin showed synergy with dexamethasone in three of six lines, with vincristine in four of six, with daunorubicin and with 4-hydroperoxycyclophosphamide in all six, and ANTAGONISM with methotrexate and with L-asparaginase in four of six.

    PMID 30679330DOI
    What this evidence cannot show

    Read from the PMC HTML full text on 2026-09-20 (efetch and Europe PMC both return an empty abstract because this is a Haematologica letter). Six cell lines, Bliss-independence scoring with Chalice software, no primary patient samples and no in-vivo arm for the combinations other than cytarabine. A cell-line synergy score is a weak predictor of clinical combination benefit.

    Evidence that argues against this (1)
    • The same panel gives 6/6 synergy for daunorubicin, so a middling dexamethasone result should be read against the panel's dynamic range rather than as a null. It is also a result about one HDAC inhibitor, and this graph carries four (romidepsin, vorinostat, panobinostat, chidamide) which are not pharmacologically equivalent. Gruber TA et al. 2026
  5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    In a Cochrane review of 32 randomised trials (4,395 preterm infants) of systemic postnatal corticosteroids started within the first six days of life, dexamethasone reduced bronchopulmonary dysplasia at 36 weeks (RR 0.72, 95% CI 0.63-0.82, high certainty) and increased cerebral palsy (RR 1.77, 95% CI 1.21-2.58, high certainty) and gastrointestinal perforation (RR 1.73, 95% CI 1.20-2.51, high certainty), whereas hydrocortisone did not increase cerebral palsy (RR 1.05, 95% CI 0.66-1.66) and did reduce mortality (RR 0.80, 95% CI 0.65-0.99).

    Early (< 7 days) systemic postnatal corticosteroids for prevention of bronchopulmonary dysplasia in preterm infants. · Doyle LW, Cheong JL, Hay S, Manley BJ, Halliday HL · Cochrane Database Syst Rev 2021
    PMID 34674229DOI
    What this evidence cannot show

    **Scope is the whole issue.** Preterm infants, many extremely preterm, treated in the first six days of life for lung disease, at BPD-literature doses and schedules, against a no-steroid comparator. This is NOT a measurement of dexamethasone in infant ALL, where the dose is 6 mg/m2/day for five weeks, the alternative is not 'no steroid', and the competing risk is death from leukaemia. Certainty is high for the question the review asks and the review does not ask this one. Cited because it is the only high-quality randomised evidence about this specific molecule in this specific age group.

    Evidence that argues against this (1)
    • The comparison that matters for leukaemia - dexamethasone against another glucocorticoid at antileukaemic doses in infants - has never been run, so the Cochrane result cannot be translated into a recommendation. Withholding or substituting the steroid in an infant with KMT2A-rearranged ALL would trade a documented neurodevelopmental risk against an undocumented change in a disease whose 6-year EFS is 46.1%. Pieters R et al. 2019
  6. E6established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.trial_registry

    The ClinicalTrials.gov structured intervention list for Interfant-06 (NCT00550992) names prednisone, prednisolone and therapeutic hydrocortisone and omits dexamethasone entirely, although the same record's detailed description prescribes dexamethasone on induction days 8-28 and in OCTADA(D).

    NCT00550992
    What this evidence cannot show

    Verified against the ClinicalTrials.gov v2 API on 2026-09-20 by counting occurrences of each drug name in `protocolSection.armsInterventionsModule.interventions` and in `protocolSection.descriptionModule.detailedDescription`. This is a defect in the registry record and not in the trial; it is recorded because the structured field is what automated drug enumerations read.

    Evidence that argues against this (1)
    • The same intervention list also omits doxorubicin, and in that case the omission is correct - Interfant-06 contains no doxorubicin anywhere, in the description or in the protocol document. So the list is unreliable in one direction only, which is worse than being uniformly wrong: absence from it carries no information either way. Interfant-06 study group / SKION (Stichting Kinderoncologie Nederland et al. 2006
  7. E7weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.trial_registry

    This beat found no randomised comparison of dexamethasone against prednisone or prednisolone in infants under 1 year with acute lymphoblastic leukaemia; Interfant-06 uses both sequentially without comparing them.

    NCT00550992
    What this evidence cannot show

    A bounded absence. Bound: PubMed 2026-09-20. Vocabularies searched: dexamethasone, prednisone, prednisolone, DEXA, glucocorticoid, corticosteroid, each combined with infant, with MLL or KMT2A, and with Interfant; `(MLL OR KMT2A) AND infant AND (prednisolone OR prednisone)` returns 44 records, all title-checked, with no steroid-randomisation trial among them; `dexamethasone versus prednisone randomized childhood acute lymphoblastic leukemia AIEOP-BFM` returns 0, and that particular zero is NOT relied on because the parser expanded 'random allocation' into unrelated MeSH sets - the 44-record title check is what bounds the claim. What would change it: an infant subgroup analysis inside one of the large paediatric steroid randomisations, which would not be reachable from these strings.

    Evidence that argues against this (1)
    • An absence of a randomised comparison in infants is not an absence of evidence about the choice. The preterm Cochrane data are randomised, high certainty, and about the same two candidate molecule classes in the same age band, and they point in one direction. That is weak evidence for the leukaemia question and it is not no evidence. Doyle LW et al. 2021
Connected to