Oncofusion

start / graph / drug-mercaptopurine

Drugmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q51-backbone

6-Mercaptopurine - two years of a nightly tablet whose lethal dose is a genotype, given as a liquid to a baby who cannot swallow it and must not have it with milk

also called 6-mercaptopurine · mercaptopurine · 6-MP · 6MP · MP · Purinethol · purine-6-thiol · thiopurine (class)

Mercaptopurine is the nightly tablet that runs for most of the two years of treatment. A small fraction of children carry a gene variant that makes the normal dose overwhelmingly toxic, and the practical problems in a baby are as much about how you get an accurate small dose into her as about the drug itself.

The same thing, for a scientist

6-Mercaptopurine is a purine antimetabolite converted by HPRT to thioinosine monophosphate and onward to thioguanine nucleotides that are incorporated into DNA, with methylthioinosine metabolites additionally inhibiting de-novo purine synthesis. Interfant-06 gives it at **three different doses in three phases** [E1]: **60 mg/m2/day orally on protocol IB days 1-28** with an explicit cumulative target of 1,680 mg/m2 that omitted doses must be made up to; **25 mg/m2/day on MARMA part I days 1-14**; and **50 mg/m2/day throughout maintenance** to 104 weeks from diagnosis. The protocol specifies administration **on an empty stomach, in the evening, and not with milk** [E1]. **The KMT2A/infant evidence is mixed and worth stating precisely:** thiopurines showed no infant-versus-older-children difference in the MTT assay, but **proB ALL of all ages was relatively more resistant to thiopurines** as well as to prednisolone and L-asparaginase [E2] - and proB is the immunophenotype of KMT2A-rearranged infant ALL. Separately, mercaptopurine is one of seven drugs with differential ex-vivo activity between children and adults, and the **only one showing subtype-independent greater sensitivity in children** [E3]. The dominant clinical variable is pharmacogenetic: TPMT and NUDT15 genotype determine tolerated dose, and this beat found **no infant-specific thiopurine dosing evidence at all** [E4].

Go deeper
Three doses, three phases, one cumulative rule
phase6-MP dosenote
protocol IB, days 1-2860 mg/m2/day POplanned cumulative 1,680 mg/m2 (28 x 60); omitted doses must be made up until that total is reached
MARMA part I, days 1-1425 mg/m2/day POthe low-dose phase, alongside high-dose methotrexate
maintenance, continuous50 mg/m2/day POwith oral methotrexate 20 mg/m2 weekly, to 104 weeks from diagnosis

Administration rules from the protocol 1: empty stomach, in the evening, not with milk. And an interlock: if a cytarabine block in protocol IB has to be postponed or interrupted, the 6-mercaptopurine is interrupted too, with the omitted doses made up to the cumulative target 1.

The cumulative-dose rule is unusual and revealing. For no other oral drug in this protocol is a total specified with an instruction to make up missed doses. It says the trialists believed 6-MP's effect is cumulative-exposure-dependent rather than schedule-dependent - a belief the maintenance thiopurine literature broadly supports and which nobody has tested in infants.

Pharmacogenetics: the one place where an individual dose is already personalised, and infants are absent from it

Thiopurine S-methyltransferase (TPMT) and NUDT15 variants cause severe, sometimes life-threatening myelosuppression at conventional doses, and pre-emptive genotype-guided dose reduction is among the best-established examples of clinical pharmacogenetics anywhere in medicine. A randomised multicentre trial of individualised mercaptopurine dosing in Chinese children with ALL exists 4.

What this beat could not find is any infant-specific thiopurine evidence - no infant TPMT/NUDT15 dosing study, no infant thiopurine PK, no infant-stratified toxicity analysis. That absence is recorded in E5 with its search bound. Genotype is of course age-independent, so the genotype-guided half transfers cleanly; what does not transfer is the starting dose it modifies, which in this protocol is a body-surface-area dose already multiplied by 2/3 or 3/4 for reasons nobody derived.

And there is a practical toxicity nobody writes papers about. 6-MP is supplied as 50 mg tablets. A 4-month-old at 50 mg/m2/day on a body surface area of roughly 0.3 m2 needs about 15 mg - a fraction of a tablet, prepared as an extemporaneous suspension, given on an empty stomach, in the evening, avoiding milk, to a patient who is milk-fed. The protocol's administration instructions 1 describe a regimen that is materially harder to deliver accurately in an infant than in a ten-year-old, and adherence and dose accuracy are the variables most strongly linked to relapse in the maintenance thiopurine literature.

Facts and figures
targetHPRT-mediated conversion to thioinosine monophosphate and onward to thioguanine nucleotides incorporated into DNA; methylthioinosine metabolites additionally inhibit de-novo purine synthesis. No HPRT1, TPMT or NUDT15 node exists in this graph as of cycle 28.
modalitysmall molecule, purine antimetabolite (thiopurine), oral
interfant06 positionprotocol IB days 1-28 at 60 mg/m2/day (cumulative target 1,680 mg/m2, omitted doses made up); MARMA part I days 1-14 at 25 mg/m2/day; maintenance continuously at 50 mg/m2/day to 104 weeks from diagnosis
administration ruleempty stomach, in the evening, not with milk
infant dose adjustmentx 2/3 under 6 months, x 3/4 from 6 through 12 months; no pharmacogenetic adjustment specified in the April 2006 protocol
cumulative dose ruleprotocol IB has a planned cumulative total of 1,680 mg/m2 and omitted doses must be made up to it - the only oral drug in the protocol with such a rule, implying the trialists treated 6-MP effect as cumulative-exposure-dependent
kmt2a specific evidenceSPLIT AND MODEST. Null on the infant-versus-older-children axis; relatively MORE resistant in proB ALL of all ages, which is the immunophenotype of KMT2A-rearranged infant ALL, but unquantified. Separately the only drug of 21 showing subtype-independent greater sensitivity in children than adults.
pharmacogeneticsTPMT and NUDT15 variants cause severe myelosuppression at conventional doses; genotype-guided dose reduction is standard practice and is absent from the April 2006 Interfant-06 protocol. Allele frequencies differ markedly by ancestry, so a dosing algorithm derived in one population does not transfer.
practical infant problemsupplied as 50 mg tablets; an infant at 50 mg/m2/day on roughly 0.3 m2 needs about 15 mg, i.e. an extemporaneous suspension, given on an empty stomach, in the evening, avoiding milk, to a milk-fed patient, nightly for up to two years
key ae
[
  "myelosuppression, severe and prolonged in TPMT or NUDT15 variant carriers",
  "hepatotoxicity including transaminitis and cholestasis",
  "nodular regenerative hyperplasia and portal hypertension (more characteristic of thioguanine - see drug-thioguanine)",
  "pancreatitis",
  "infection during prolonged maintenance"
]
resistance
[
  "high TPMT activity shunting drug to inactive methylated metabolites",
  "loss of HPRT preventing activation",
  "mismatch repair deficiency preventing thioguanine-nucleotide-triggered apoptosis - a documented route to acquired thiopurine resistance at relapse",
  "no infant- or KMT2A-specific mechanism identified; the proB relative resistance in PMID 14712291 is unquantified"
]
trials
[
  "NCT00550992"
]
How this could bite us back
  • **A homozygous TPMT- or NUDT15-deficient infant given the protocol dose will develop profound, prolonged marrow aplasia, and the April 2006 protocol contains no genotyping requirement.** The test is cheap, the genotype does not change with age, and the consequence of missing it is measured in weeks of aplasia in a patient whose induction deaths are already infectious.
  • Two years of a nightly oral drug in an infant is an adherence and dose-accuracy problem before it is a pharmacology problem. The protocol requires an empty stomach, an evening dose and no milk, in a patient who feeds on milk through the night.
  • The protocol treats 6-MP as cumulative-exposure-dependent and requires missed doses to be made up. If that belief is wrong and schedule matters, the make-up rule delivers concentrated exposure at the wrong time; if it is right, the two-thirds age multiplier is removing a fifth of a cumulative target that the same protocol elsewhere insists must be reached.
  • ProB ALL - this disease's immunophenotype - is relatively thiopurine-resistant in the only study that looked, alongside being steroid- and asparaginase-resistant. That makes 6-MP the third of four backbone agents to which this phenotype is relatively resistant, with cytarabine the sole exception.
  • Mismatch-repair deficiency is a documented route to acquired thiopurine resistance at relapse, which means two years of maintenance can select for a mutator phenotype in the cells that survive. In a disease that relapses early and switches lineage, the selection pressure applied for longest is the one least examined.
Still open
  1. What thioguanine-nucleotide exposure does an infant achieve from 50 mg/m2/day of an extemporaneous suspension, multiplied by 2/3? The pharmacodynamic assay is routine in adult gastroenterology and has apparently never been applied in infant ALL maintenance.
  2. Should Interfant successors require TPMT and NUDT15 genotyping? The April 2006 protocol does not, genotype is age-invariant, and severe myelosuppression during two years of maintenance in an infant is a life-threatening event.
  3. Is the relative thiopurine resistance of proB ALL real and how large is it? It is reported qualitatively in one 2004 MTT study, it applies to the immunophenotype of this disease, and nobody has followed it up.
  4. Does dose accuracy from extemporaneous liquid preparation vary enough to matter? Adherence and dose accuracy are the best-documented modifiable determinants of relapse in ALL maintenance, and the infant delivery problem is the worst in paediatrics.
Every source behind this entry (5)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Interfant-06 gives 6-mercaptopurine 60 mg/m2/day orally on protocol IB days 1-28 with a planned cumulative total of 1,680 mg/m2 that omitted doses must be made up to, 25 mg/m2/day on MARMA part I days 1-14, and 50 mg/m2/day through maintenance with oral methotrexate 20 mg/m2 weekly to 104 weeks from diagnosis; it is to be given on an empty stomach, in the evening, and not with milk; if a cytarabine block in protocol IB is postponed or interrupted the 6-mercaptopurine is interrupted as well; the body-surface-area dose is multiplied by 2/3 under 6 months and 3/4 from 6 through 12 months.

    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. The protocol gives no TPMT or NUDT15 genotyping requirement and no pharmacogenetic dose-adjustment rule, which is unsurprising for an April 2006 document but means Interfant-06 patients were dosed without the adjustment that is now standard.

    Evidence that argues against this (1)
    • A protocol dose is not a delivered dose for an oral drug taken at home for two years. Adherence is the best-documented modifiable determinant of relapse in ALL maintenance, and in an infant the dose must be prepared as an extemporaneous liquid from a 50 mg tablet, on an empty stomach, avoiding milk, in a milk-fed patient. Zhou Y et al. 2024
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In the MTT assay, thiopurines showed no difference in resistance between 44 infants with ALL and 425 children over 1 year with common/preB ALL, but proB ALL of all ages was relatively more resistant to thiopurines as well as to L-asparaginase and prednisolone compared with common/preB ALL.

    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 2004
    PMID 14712291DOI
    What this evidence cannot show

    The thiopurine result is split: null on the infant axis, positive on the immunophenotype axis. ProB is the characteristic immunophenotype of KMT2A-rearranged infant ALL, so the immunophenotype finding is the one that bears on this disease, and the abstract gives no fold-difference or P value for the thiopurine limb of the proB comparison - only the direction. A short-term MTT is a weak assay for a drug whose mechanism requires DNA incorporation over days.

    Evidence that argues against this (1)
    • The relative thiopurine resistance of proB ALL was not separately quantified and sits alongside the far larger prednisolone and asparaginase effects in the same comparison, so it could be a small effect reported qualitatively. A 2004 MTT measurement also cannot speak to thioguanine-nucleotide incorporation, which is the pharmacodynamic endpoint the thiopurine field now uses. Yoshimura S et al. 2024
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Across 767 children and 309 adults with newly diagnosed B-ALL profiled for ex-vivo sensitivity to 21 drugs, mercaptopurine was one of seven drugs showing differential activity between children and adults and was the one that exhibited subtype-INDEPENDENT greater sensitivity in children, the other six being explained by age-related differences in leukaemia molecular subtype.

    Impact of Age on Pharmacogenomics and Treatment Outcomes of B-Cell Acute Lymphoblastic Leukemia. · Yoshimura S, Li Z, Gocho Y, Yang W, Crews KR, Lee SHR, et al. · J Clin Oncol 2024
    PMID 39102629DOI
    What this evidence cannot show

    Children versus adults, not infants versus older children; the youngest patients are not separately analysed in the abstract. 'Subtype-independent' means the effect survived adjustment for the 23 molecular subtypes identified, which is a statement about that adjustment and not proof of an age-intrinsic mechanism. Ex-vivo sensitivity of primary blasts over a short culture is an indirect assay for a thiopurine.

    Evidence that argues against this (1)
    • A subtype-independent age effect for mercaptopurine could reflect host pharmacogenetics rather than blast biology, since the same cohorts differ in TPMT and NUDT15 allele frequency by ancestry and in achieved thioguanine-nucleotide exposure. An ex-vivo blast assay cannot distinguish an intrinsically more sensitive blast from a population difference in the host. Zhou Y et al. 2024
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Individualised mercaptopurine dosing has been tested in a multicentre randomised controlled trial in Chinese children with acute lymphoblastic leukaemia.

    Individualized Use of 6-Mercaptopurine in Chinese Children with ALL: A Multicenter Randomized Controlled Trial. · Zhou Y, Wang L, Sun LR, Zhang L, Wang HM, Liu XT, et al. · Clin Pharmacol Ther 2024
    PMID 37753808DOI
    What this evidence cannot show

    Cited at title level for the existence and design of randomised individualised thiopurine dosing in children; this beat did not obtain the full text and states no effect size, no dosing algorithm and no outcome. It is a paediatric ALL trial and not an infant trial, and allele frequencies for NUDT15 differ substantially between the Chinese population studied and the European populations of the Interfant trials, so its dosing algorithm would not transfer unmodified.

    Evidence that argues against this (1)
    • Individualisation of a maintenance thiopurine dose targets tolerability and thioguanine-nucleotide exposure, both surrogates. Whether genotype-guided dosing improves event-free survival, as opposed to reducing toxicity and interruptions, remains the open question in the field and is not settled by a single national trial. Yoshimura S et al. 2024
  5. E5weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

    This beat found no infant-specific thiopurine dosing, pharmacokinetic or pharmacogenetic study - no TPMT or NUDT15 dosing analysis restricted to or stratified for children under 1 year, and no infant mercaptopurine pharmacokinetic measurement.

    Effect of developmental changes on pharmacokinetics of drugs used in the treatment of infant acute lymphoblastic leukaemia-A comprehensive review. · de Kluis T, Brigitha LJ, Hanff LM, Huitema ADR, van der Sluis IM, Diekstra MHM · Br J Clin Pharmacol 2025
    PMID 41139514DOI
    What this evidence cannot show

    A bounded absence, weakly bounded, and recorded as weak for that reason. Bound: PubMed 2026-09-20. Strings searched: `NUDT15 TPMT thiopurine dosing children acute lymphoblastic leukemia mercaptopurine` (19 records, all title-checked, none infant-restricted); `(MLL OR KMT2A) AND infant AND (prednisolone OR prednisone)` (44 records, title-checked, no thiopurine study); vocabularies covered 6-MP, 6MP, mercaptopurine, thiopurine, TPMT, NUDT15. NOT searched: 'purinethol', non-English literature, paediatric formulary and extemporaneous-preparation literature, which is where an infant dosing practice would most plausibly be documented. What would change it: an infant subgroup inside a national pharmacogenetics cohort, which these strings would not reach.

    Evidence that argues against this (1)
    • TPMT and NUDT15 genotype is age-invariant, so the pharmacogenetic half of thiopurine dosing transfers to infants without needing infant-specific evidence. The absence recorded here concerns the starting dose that genotype modifies and the drug exposure achieved from an extemporaneous liquid, not the genotype-guided adjustment itself. Zhou Y et al. 2024
Connected to