Cytarabine - the one conventional drug KMT2A-rearranged infant ALL is MORE sensitive to, and the only case in this beat where a mechanism changed a protocol
also called cytarabine · ara-C · araC · Ara-C · cytosine arabinoside · cytarabine arabinoside · HD-AraC · HDAC (high-dose ara-C, ambiguous - see caveats) · Cytosar · 1-beta-D-arabinofuranosylcytosine
Cytarabine is the one ordinary chemotherapy drug that infant leukaemia cells are unusually good at taking up and being killed by - the opposite of the steroid story. That laboratory finding is the reason cytarabine appears so heavily in the protocols these babies are actually given.
The same thing, for a scientist
Cytarabine is a deoxycytidine analogue that must be transported into the cell (largely by hENT1/SLC29A1) and phosphorylated by deoxycytidine kinase to ara-CTP, which is incorporated into DNA and terminates chain elongation. **In infant ALL it is the exception to the resistant phenotype.** Infant blasts are 1.9-fold more sensitive to cytarabine than older children with common/preB ALL in the MTT assay [E1] and 3.3-fold more sensitive with 2.3-fold greater ara-CTP accumulation in a dedicated study [E2]; the mechanism is 2.5-fold higher hENT1 mRNA (despite 2-fold *lower* dCK), and hENT1 expression correlates inversely with in-vitro ara-C resistance (rs = -0.58, P = 0.006) [E2]. The same sensitivity is seen in 11q23-rearranged childhood ALL and, with doxorubicin, in t(9;11) AML [E3]. **This is the only drug in this beat where the causal chain from mechanism to protocol is explicit in the literature**: Stumpel et al. state that the nucleoside-analogue observations 'led to the inclusion of cytarabine into the treatment regimens currently used for infants with ALL' [E4], and JPLSG MLL-10 added high-dose cytarabine in early intensification for KMT2A-rearranged patients and reached a 3-year EFS of 66.2% [E5]. In Interfant-06 cytarabine appears in **six** places [E6].
Where it sits in the backbone - the most-used drug in the protocol
From the registered protocol 6, cytarabine appears at more points than any other agent:
| phase | cytarabine |
|---|---|
| remission induction | IV over 30 min, days 8-21; IT day 15 |
| protocol IB (arm I) | IV days 3-6, 10-13, 17-20, 24-27; IT day 10 |
| ADE (arm II) | IV every 12 h, days 1-10 |
| MAE (arm II) | IV every 12 h, days 1-10 |
| MARMA part II | high-dose, IV over 3 h twice daily at 12-h intervals, days 15, 16, 22, 23 |
| OCTADA(D) | IV days 2-5, 9-12, 16-19, 23-26 (part I) and 37-40, 45-48 (part II); IT days 1 and 15 |
| maintenance | IT week 8 |
Both arms of the central randomisation are cytarabine-containing; the randomisation is about what is added to cytarabine, not about cytarabine.
An alias trap worth naming. HDAC is used in the paediatric-leukaemia literature for high-dose ara-C and in the chromatin literature for histone deacetylase, and this graph contains both literatures - it carries drug-romidepsin, drug-vorinostat, drug-panobinostat and drug-chidamide as HDAC inhibitors. Anything in this graph matching HDAC must be disambiguated by context before it is counted.
The mechanism, and why it runs the opposite way to intuition
Infant/MLL-rearranged blasts express less deoxycytidine kinase, the activating enzyme - which should make them more resistant - and more hENT1, the influx transporter 2. Transport wins: ara-CTP accumulation is 2.3-fold higher and hENT1 mRNA, not dCK mRNA, is what correlates with the resistance measurement 2. An independent paediatric cohort confirmed the underexpression of ara-C-metabolising enzyme genes in MLL-rearranged ALL 7, so the two halves of the paradox have been seen twice.
The practical reading is that the rate-limiting step for cytarabine in this disease is getting in, not getting activated, which is unusual - in adult AML, dCK and CDA are usually the variables of interest. It also predicts that anything reducing hENT1 would cost more here than elsewhere, and hENT1 is the transporter FLT3 signalling has been reported to modulate in paediatric acute leukaemia 8 - relevant because KMT2A-rearranged ALL overexpresses FLT3.
What was built on it, and where it stopped
It changed practice. Cytarabine's place in infant ALL regimens is attributed by the Rotterdam group to exactly these in-vitro observations 4. COG AALL0631 and then JPLSG MLL-10 added high-dose cytarabine in early intensification for KMT2A-rearranged infants; MLL-10 reported 3-year EFS 66.2% for KMT2A-rearranged (IR+HR), 94.4% for intermediate risk, 56.6% for high risk, and was able to remove transplant from a subset 5. That is the best KMT2A-rearranged infant EFS in this graph's trial entries.
And it stopped short of a cure. Stumpel et al. note that despite the cytarabine strategy, survival for MLL-rearranged infant ALL 'do[es] still not exceed 30-40%' 4, and went looking for a better nucleoside analogue - clofarabine, median LC50 ~25 nM on primary MLL-rearranged infant cells, synergistic with cytarabine 4.
And it is not uniform across patients. In three KMT2A-rearranged infant ALL xenografts, low-dose cytarabine gave a significant survival benefit in two (PER-785, MLL-5/t(10;11)) and not in the third (MLL-14/t(11;19)), a difference the authors validated in vitro and describe as representing known inter-patient variability 9. A partner-dependent difference in cytarabine response is exactly what E3 reported for 11q23 subtypes, so this is now seen from two directions.
| target | hENT1/SLC29A1-mediated uptake then deoxycytidine kinase activation to ara-CTP, which is incorporated into DNA and terminates chain elongation. No gene node for SLC29A1 or DCK exists in this graph as of cycle 28. |
|---|---|
| modality | small molecule, pyrimidine (deoxycytidine) nucleoside analogue antimetabolite |
| interfant06 position | induction IV days 8-21 and IT day 15; protocol IB IV days 3-6/10-13/17-20/24-27 and IT day 10; ADE and MAE IV q12h days 1-10; MARMA part II HIGH DOSE IV q12h days 15/16/22/23; OCTADA(D) IV days 2-5/9-12/16-19/23-26 and 37-40/45-48, IT days 1 and 15; maintenance IT week 8 |
| kmt2a specific evidence | GENUINE AND ACTED UPON. The only backbone drug whose KMT2A/infant-specific in-vitro finding is credited with changing the protocol. Infant blasts 1.9- to 3.3-fold more sensitive; 2.3-fold higher ara-CTP accumulation; 2.5-fold higher hENT1 mRNA; 11q23 ALL and t(9;11) AML both more cytarabine-sensitive in an independent assay. |
| sensitivity mechanism | elevated hENT1/SLC29A1 influx transport, NOT elevated activation - dCK mRNA is 2-fold LOWER in infants and ara-C-metabolising enzyme genes are underexpressed in MLL-rearranged ALL |
| hent1 mrna vs resistance spearman | -0.58 |
| partner dependence | t(9;11) AML most cytarabine-sensitive among 11q23 subtypes (PMID 15813846); among infant ALL xenografts, t(11;19)/MLL-14 did not benefit from low-dose cytarabine while t(10;11)/MLL-5 did (PMID 30679330) |
| alias collision warning | HDAC means high-dose ara-C in the paediatric leukaemia literature and histone deacetylase in the chromatin literature. This graph contains both. Disambiguate before counting. |
key ae
[ "myelosuppression", "mucositis", "conjunctivitis and keratitis at high dose (steroid eye drops are given prophylactically)", "cerebellar toxicity at high dose", "ara-C syndrome (fever, myalgia, rash)", "chemical arachnoiditis via the intrathecal route" ]
resistance
[ "loss or downregulation of hENT1/SLC29A1 (the influx step that carries sensitivity in this disease)", "low deoxycytidine kinase activity", "high cytidine deaminase (deactivation)", "inter-patient and partner-dependent variability demonstrated in xenografts" ]
trials
[ "NCT00550992" ]
related nucleoside analogues in this disease
[ "cladribine/2-CdA (1.7-fold more sensitive in infants)", "clofarabine (median LC50 ~25 nM on primary cells, synergistic with cytarabine)", "decitabine and zebularine (demethylating, reported active in vitro)" ]
- **Cytarabine is already given at more timepoints than any other drug in this protocol, so 'exploit the sensitivity' mostly means intensify something an infant is already saturated with.** High-dose cytarabine brings cerebellar toxicity and keratitis into a patient who cannot report either.
- The controlled test of intensified cytarabine-containing consolidation in this disease - Interfant-06's central randomisation - was negative. A mechanism that is real in vitro has already failed once to convert into a randomised gain, and the strongest supporting clinical data (MLL-10) are single-arm.
- The sensitivity is carried by an influx transporter, which means it is a property that can be lost by downregulating one gene. Any selective pressure that reduces hENT1 removes the one conventional advantage this disease offers, and relapse samples have not been checked for it.
- Cytarabine is a myeloid-directed agent used in a lymphoid disease that switches lineage under pressure. KMT2A-rearranged infant leukaemia is the archetype of lineage switch, and a cytarabine-heavy backbone selects in the direction of the compartment that escapes CD19-directed therapy.
- The most attractive partner for cytarabine found in this disease - an HDAC inhibitor - antagonises methotrexate and asparaginase in four of six infant cell lines. Adding it means choosing which phase of a protocol to protect, and the phases it would damage are the ones that deliver CNS control.
- Does hENT1 protein or transport flux, rather than mRNA, predict cytarabine response in KMT2A-rearranged infants? The entire mechanism rests on a transcript measurement correlated with an LC50, and hENT1 is a membrane protein whose activity is regulated post-transcriptionally.
- Why do some KMT2A-rearranged xenografts not respond to cytarabine at all, and does the partner gene predict it? Two independent lines of evidence (t(9;11) in AML, t(11;19) in the xenograft panel) point at partner-dependence, and nobody has tested hENT1 across a partner panel.
- If FLT3 signalling raises hENT1, does FLT3 inhibition antagonise cytarabine in this disease? AALL0631 gave lestaurtinib on a cytarabine backbone and found no benefit; the antagonism hypothesis is testable in the stored correlative samples.
- Clofarabine is 40-fold more potent than cytarabine on these cells in vitro and synergises with it, and no infant trial has tested it up front. What is the toxicity ceiling in a 4-month-old?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In MTT assays, 44 infants with ALL were 1.9-fold MORE sensitive to cytarabine than 425 children over 1 year with common/preB ALL (P = 0.001), and 1.7-fold more sensitive to 2-chlorodeoxyadenosine (P < 0.001), while being >580-fold resistant to prednisolone and 12-fold resistant to L-asparaginase in the same assay.
In vitro drug-resistance profile in infant acute lymphoblastic leukemia in relation to age, MLL rearrangements and immunophenotype. · Ramakers-van Woerden NL, Beverloo HB, Veerman AJ, Camitta BM, Loonen AH, van Wering ER, et al. · Leukemia 2004What this evidence cannot show
A 1.9-fold LC50 difference is modest beside the steroid figures in the same table and should not be inflated into a claim of cytarabine dependence. The authors attribute much of the profile to the proB immunophenotype rather than to MLL rearrangement, and proB ALL of all ages showed the same cytarabine sensitivity. MTT LC50 on a bulk population cannot report on a subclone.
Evidence that argues against this (1)
- The same study found NO difference between infants and older children for vincristine, anthracyclines, thiopurines or epipodophyllotoxins, so the in-vitro profile is not a general statement that infant blasts behave unusually - only three drug classes moved, and cytarabine moved least of the three. Ramakers-van Woerden NL et al. 2004
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Infant ALL cells were 3.3-fold more sensitive to ara-C (P = 0.007) and accumulated 2.3-fold more ara-CTP (P = 0.011) than older children; infants expressed 2-fold less deoxycytidine kinase mRNA (P = 0.026) but 2.5-fold more hENT1 mRNA (P = 0.001), hENT1 expression correlated inversely with in-vitro ara-C resistance (rs = -0.58, P = 0.006), and the same dCK and hENT1 differences separated MLL-rearranged (n = 14) from germline-MLL (n = 25) cases.
Differential mRNA expression of Ara-C-metabolizing enzymes explains Ara-C sensitivity in MLL gene-rearranged infant acute lymphoblastic leukemia. · Stam RW, den Boer ML, Meijerink JP, Ebus ME, Peters GJ, Noordhuis P, et al. · Blood 2003What this evidence cannot show
n = 18 infants and 24 older children; mRNA by TaqMan plus an Affymetrix screen, with no protein-level or transport-flux measurement of hENT1 itself. Correlation of mRNA with an LC50 across a mixed cohort is not a demonstration that transport is rate-limiting in any individual patient. The dCK result points the opposite way to the phenotype and is not explained.
Evidence that argues against this (1)
- An independent paediatric cohort found ara-C-metabolising enzyme genes UNDEREXPRESSED in MLL-rearranged ALL, which reproduces the dCK half of this result and is, read alone, a prediction of ara-C RESISTANCE. The two studies agree on the measurement and the sensitivity phenotype has to be carried by transport rather than by activation for both to be true. Mata JF et al. 2006
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In a fluorometric microculture cytotoxicity assay of 132 children with AML and 178 with ALL, children with ALL and an 11q23 rearrangement (n = 22) were significantly more sensitive to cytarabine than children without (n = 156, P = 0.026) even after stratifying for white blood cell count, and in AML those with t(9;11) (n = 10) were more sensitive to both cytarabine (P < 0.001) and doxorubicin (P = 0.005) than non-11q23 cases; the authors conclude that cellular drug resistance correlates with both cell lineage and the type of 11q23 rearrangement.
Cellular drug sensitivity in MLL-rearranged childhood acute leukaemia is correlated to partner genes and cell lineage. · Palle J, Frost BM, Forestier E, Gustafsson G, Nygren P, Hellebostad M, et al. · Br J Haematol 2005What this evidence cannot show
Small subgroups (n = 10 for t(9;11), n = 22 for 11q23 ALL) and a different assay (FMCA) from the MTT work, so the fold-differences are not comparable across studies. Partner-gene subgroups were not individually powered; the partner-dependence conclusion rests on t(9;11) separating from 'other 11q23' as a pooled category.
Evidence that argues against this (1)
- This study also reports an 'AML-profile' of relative resistance to glucocorticoids and vincristine common to all three groups, so the lineage of the leukaemia, not the 11q23 rearrangement, accounts for part of what looks like a rearrangement effect. The cytarabine finding in ALL survives stratification for WBC but was not stratified for immunophenotype. Ramakers-van Woerden NL et al. 2004
- E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
MLL-rearranged infant ALL responds well in vitro to nucleoside analogues, and these observations led to the inclusion of cytarabine into the treatment regimens currently used for infants with ALL; nonetheless survival for MLL-rearranged infant ALL still does not exceed 30-40%, and clofarabine targets primary MLL-rearranged infant cells at median LC50 ~25 nM with synergy in combination with cytarabine.
The potential of clofarabine in MLL-rearranged infant acute lymphoblastic leukaemia. · Stumpel DJ, Schneider P, Pieters R, Stam RW · Eur J Cancer 2015What this evidence cannot show
The attribution of protocol design to the in-vitro findings is the authors' own account of their field's history, not a documented decision record, and the Rotterdam group authored both the in-vitro work and this statement. The 30-40% figure is a contemporaneous summary, and MLL-10 subsequently reported 66.2% 3-year EFS in KMT2A-rearranged infants. The clofarabine data are in vitro only.
Evidence that argues against this (1)
- If cytarabine's inclusion followed from in-vitro sensitivity, the same logic should have delivered a larger clinical gain than it did. Interfant-06's whole-cohort 6-year EFS of 46.1% did not improve on Interfant-99 despite a cytarabine-heavy backbone in both arms, and the randomisation between a cytarabine/anthracycline/etoposide myeloid-style consolidation and a lymphoid-style one was negative (P = .47). Pieters R et al. 2019
- E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In JPLSG MLL-10, COG AALL0631-modified chemotherapy with the addition of high-dose cytarabine in early intensification was given to KMT2A-rearranged infants and transplant was restricted to the high-risk group; 3-year EFS was 66.2% (SE 5.6%) for KMT2A-rearranged patients, 94.4% (SE 5.4%) for intermediate risk and 56.6% (SE 6.8%) for high risk, and the approach eliminated transplant for a subset of KMT2A-rearranged infants.
A risk-stratified therapy for infants with acute lymphoblastic leukemia: a report from the JPLSG MLL-10 trial. · Tomizawa D, Miyamura T, Imamura T, Watanabe T, Moriya Saito A, Ogawa A, et al. · Blood 2020What this evidence cannot show
Single-arm, 90 eligible infants, no randomisation of the high-dose cytarabine addition, so the cytarabine contribution cannot be separated from the concurrent change in transplant policy, the risk stratification, or from cohort differences against historical controls. Standard errors around 6% on subgroups of 19 and 56. A Japanese national cohort compared with international trials.
Evidence that argues against this (1)
- The Interfant-06 randomisation is the controlled test of intensified myeloid-style cytarabine-containing consolidation in this disease and it was negative: 6-year DFS 39.3% for ADE+MAE versus 36.8% for protocol IB, P = .47. MLL-10's better EFS therefore cannot be attributed to the cytarabine intensification without assuming the two cohorts are comparable, which they are not. Pieters R et al. 2019
- E6established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.trial_registry
Interfant-06 gives cytarabine in induction (IV days 8-21, IT day 15), protocol IB (IV days 3-6, 10-13, 17-20, 24-27; IT day 10), ADE and MAE (IV every 12 hours days 1-10), MARMA part II (high dose, IV over 3 hours twice daily at 12-hour intervals on days 15, 16, 22, 23), OCTADA(D) (IV days 2-5, 9-12, 16-19, 23-26 and days 37-40, 45-48; IT days 1 and 15) and maintenance (IT week 8) - more separate administrations than any other agent in the protocol.
Interfant-06: International Collaborative Treatment Protocol for Infants Under One Year With Acute Lymphoblastic or Biphenotypic Leukemia (NCT00550992) · ClinicalTrials.gov 2007What this evidence cannot show
Registry record fetched 2026-09-20 from the ClinicalTrials.gov v2 API, not the protocol document. No doses are stated for any agent, so 'high-dose' here is the registry's own wording and this entry does not supply a mg/m2. Arm II phases apply only to medium- and high-risk patients randomised to the experimental arm.
Evidence that argues against this (1)
- Appearing in the most phases is not the same as contributing the most. The protocol's one randomised comparison was between two cytarabine-containing consolidations and was negative, which means the protocol contains no internal evidence about how much of the outcome cytarabine carries. Pieters R et al. 2019
- E7moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Cytosine arabinoside-metabolising enzyme genes are underexpressed in children with MLL gene-rearranged acute lymphoblastic leukaemia.
Cytosine arabinoside-metabolizing enzyme genes are underexpressed in children with MLL gene-rearranged acute lymphoblastic leukemia. · Mata JF, Scrideli CA, Queiroz RP, Mori BO, Emerenciano M, Pombo-de-Oliveira MS, et al. · Braz J Med Biol Res 2006What this evidence cannot show
Cited for the direction of the enzyme finding only; this beat read the title and did not obtain the full text, so cohort size, which enzymes, and effect sizes are not stated here. It is used as independent corroboration that the activation machinery is low in this subtype, which is the counter-intuitive half of E2, and not for any quantitative claim.
Evidence that argues against this (1)
- Underexpressed activating enzymes predict cytarabine resistance, and the observed phenotype is sensitivity. Taken alone this paper would license the wrong conclusion; it is only interpretable alongside the transporter measurement in E2. Stam RW et al. 2003
- E8weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper
FLT3 has been implicated in cytarabine transport by human equilibrative nucleoside transporter 1 in paediatric acute leukaemia.
FLT3 is implicated in cytarabine transport by human equilibrative nucleoside transporter 1 in pediatric acute leukemia. · Català A, Pastor-Anglada M, Caviedes-Cárdenas L, Malatesta R, Rives S, Vega-García N, et al. · Oncotarget 2016What this evidence cannot show
Title-level citation; this beat did not read the full text and states no effect size, direction or cohort. It is recorded because KMT2A-rearranged ALL overexpresses FLT3 and hENT1 is the transporter that carries cytarabine's sensitivity in this disease, so an FLT3-hENT1 link is a testable bridge between two things the graph already holds - not because the link is established.
Evidence that argues against this (1)
- If FLT3 signalling raised hENT1 and thereby cytarabine sensitivity, FLT3 inhibition would be predicted to ANTAGONISE cytarabine in KMT2A-rearranged infant ALL. COG AALL0631 added the FLT3 inhibitor lestaurtinib to a cytarabine-containing backbone and found no overall EFS benefit (36 +/- 6% with vs 39 +/- 7% without, P = 0.67), which is compatible with antagonism but does not demonstrate it. Brown PA et al. 2021
- E9moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
In three KMT2A-rearranged infant ALL xenograft models, low-dose cytarabine alone gave a significant survival benefit for PER-785 and MLL-5 (t(10;11)) but not for MLL-14 (t(11;19)); the differential sensitivity was validated in vitro, and combining romidepsin with low-dose cytarabine produced a profound, highly significant increase in median survival in all three models.
Romidepsin enhances the efficacy of cytarabine in vivo, revealing histone deacetylase inhibition as a promising therapeutic strategy for KMT2A-rearranged infant acute lymphoblastic leukemia. · Cheung LC, Cruickshank MN, Hughes AM, Singh S, Chua GA, Ford J, et al. · Haematologica 2019What this evidence cannot show
Read from the PMC HTML full text on 2026-09-20 after efetch and Europe PMC both returned an empty abstract - the abstract field is empty because this is a Haematologica letter. Three xenograft models, 7-8 mice per group; 'profound' is the authors' word for a median-survival difference by log-rank, not a cure. Xenografts in immune-deficient mice do not model the infection toxicity that limits intensification in infants.
Evidence that argues against this (1)
- The same paper reports that romidepsin ANTAGONISED methotrexate and L-asparaginase in four of six infant ALL cell lines, so the proposal that follows from the cytarabine synergy - schedule romidepsin into protocol IB and the second half of OCTADAD - is constrained by phases where the combination would be expected to do harm. A synergy result in one phase is not a licence for the backbone. Cheung LC et al. 2019
- part of →Interfant-06 — 651 infants, and the definitive answer that myeloid-style consolidation does not help
induction, protocol IB, ADE, MAE, MARMA part II (high dose), OCTADA(D), maintenance IT - the most-used agent in the protocol
present in six phases
infant blasts 1.9-3.3 fold MORE sensitive than older children's
11q23-rearranged ALL and t(9;11) AML both more cytarabine-sensitive than non-rearranged
romidepsin + low-dose cytarabine gave a profound survival increase in three KMT2A-r infant ALL xenografts
FLT3 implicated in hENT1-mediated cytarabine transport; KMT2A-r ALL overexpresses FLT3 - a testable antagonism
the two poles of the infant in-vitro profile in one assay: >580-fold resistant to prednisolone, 1.9-fold more sensitive to cytarabine
the two halves of MARMA: HD-MTX in part I, HD-cytarabine in part II
the two cytotoxics of protocol IB, interlocked by protocol rule with mercaptopurine
- ← confers resistance toLEPR - the marker that defines the SCF-supplying perivascular niche cell, the receptor for an adipocyte signal that blunts cytarabine, and a gene_exact query that returns the wrong protein first
leptin-LEPR signalling blunts cytarabine cytotoxicity; the LEPR antagonist Allo-aca restores it
both are activated by deoxycytidine kinase, so loss of that enzyme confers resistance to both