Treosulfan - a European conditioning standard for three decades that the FDA first approved in January 2025, for a diagnosis and an age that exclude the infant with KMT2A-rearranged ALL
also called treosulfan · treosulphan · dihydroxybusulfan · L-threitol 1,4-bismethanesulfonate · TREO · Trecondi (EU trade name) · Ovastat (older EU trade name) · Grafapex (US trade name, approved 2025) · prodrug of diepoxybutane (class: bifunctional alkylsulfonate prodrug)
Treosulfan is a close chemical relative of busulfan that works the same way but is gentler on the liver, and European transplant units have used it in children for years. The American regulator only approved it in January 2025, and only for a different leukaemia and only from one year of age - so a six-month-old with this disease is outside its licence on both counts.
The same thing, for a scientist
Treosulfan is a bifunctional alkylsulfonate PRODRUG that converts non-enzymatically, at physiological pH, into a monoepoxide and then diepoxybutane, which alkylate DNA. Within the FORUM trial's prespecified comparison, treosulfan/thiotepa/fludarabine and busulfan/thiotepa/fludarabine gave statistically indistinguishable 3-year overall survival (0.72 vs 0.71), event-free survival (0.55 vs 0.60), relapse, non-relapse mortality and GVHD in 308 children with ALL, with one fatal veno-occlusive death in each arm [E1]. Unlike busulfan it is used without routine therapeutic drug monitoring, though a prospective paediatric trial has proposed a target cumulative AUC of 4,800 mg*h/L and found mortality rising 1.46-fold per additional 1,000 mg*h/L [E2]. The US approval came on 2025-01-21 as a Type 1 New Molecular Entity, NDA 214759, restricted to AML and MDS in patients one year of age and older, in combination with fludarabine [E3].
The regulatory geography, which is the whole reason this drug is invisible from a US vantage point1 deeper
Checked live against the openFDA Drugs@FDA and drug-label APIs on 2026-09-20 3:
| field | value |
|---|---|
| application | NDA 214759, Medexus Pharma |
| brand | GRAFAPEX |
| original submission | 2025-01-21, classified Type 1 - New Molecular Entity |
| indication | allogeneic HSCT preparative regimen, in combination with fludarabine, for AML and for MDS |
| age | 1 year of age and older |
| dose | 10 g/m2/day i.v. over 2 h on days -4, -3, -2 |
Type 1 - New Molecular Entity, in 2025, for a compound in routine European paediatric transplant practice since the 1990s. An infant under one year with KMT2A-rearranged ALL is outside this label twice: wrong diagnosis and wrong age.
This is not a statement that treosulfan is unusable in such a child - off-label conditioning is routine, and the Chiesa cohort describes treosulfan as 'given off-label in pediatric allogeneic hematopoietic stem cell transplant' 2. It is a statement about what evidence and what infrastructure exist. A drug outside its licence has no paediatric pharmacovigilance commitment attached to it, no required paediatric investigation plan for that population, and no labelled dose in the age band that matters here.
Correcting a premise in this beat's dispatch
The dispatch brief warned that 'treosulfan is used in Europe and barely in the US, so a PubMed-only search will understate it.' The clinical half of that is right and the retrieval half is wrong. Measured on PubMed on 2026-09-20:
| string | [tiab] records |
|---|---|
treosulfan OR treosulphan OR dihydroxybusulfan | 558 |
Ovastat | 8 |
Trecondi | 1 |
treosulfan AND conditioning AND children AND leukemia | 32 |
PubMed indexes treosulfan perfectly well under its generic name, in both spellings, and the trade names contribute nothing. The understatement is in practice and regulation, not in retrieval - and the two failure modes need different remedies. Rules 71 and 72 direct a searcher to trade names and class names; here that would have returned nine extra records and missed the actual asymmetry entirely, which was only visible in a regulatory database. When a drug is geographically split, check the approval registry, not just the literature index.
Busulfan or treosulfan: a prespecified comparison that found nothing
FORUM randomised TBI against chemoconditioning; within the chemoconditioning arm it prespecified a comparison of the two alkylator backbones, driven by the fact that TBI is simply unavailable in some countries. 180 patients received BU/THIO/FLU and 128 TREO/THIO/FLU, median follow-up 4.2 years 1.
| endpoint at 3 years | busulfan | treosulfan |
|---|---|---|
| overall survival | 0.71 (0.64-0.77) | 0.72 (0.63-0.79) |
| event-free survival | 0.60 (0.53-0.67) | 0.55 (0.46-0.63) |
| relapse | 0.31 (0.25-0.38) | 0.36 (0.27-0.44) |
| non-relapse mortality | 0.08 (0.05-0.13) | 0.09 (0.05-0.15) |
| GVHD-free relapse-free survival | 0.48 (0.41-0.55) | 0.45 (0.37-0.54) |
| fatal veno-occlusive disease | 1 case | 1 case |
No difference on any endpoint. The honest reading is not that the agents are equivalent - this was a non-randomised comparison inside a randomised trial, 308 patients, and confidence intervals on EFS that admit a five-point gap in either direction. The reading is that there is no evidence to prefer one, and the choice can therefore be made on availability, on monitoring infrastructure, and on the hepatic-toxicity argument that motivated treosulfan in the first place.
And the entire comparison is in patients aged over 4 years, because that is FORUM's randomisation floor. Neither agent has been compared in an infant.
The monitoring asymmetry, and why 'no TDM needed' is a claim rather than a property
Treosulfan is conventionally given at a fixed body-surface-area dose without exposure monitoring, and the 2021 review of irradiation-free conditioning states that it 'leads to comparable outcomes to busulfan-based conditioning in paediatric ALL, without the need for TDM to date' 4. That phrasing is careful, and the caution is warranted.
In 87 children given treosulfan-fludarabine conditioning prospectively, each additional 1,000 mgh/L of treosulfan AUC raised the mortality hazard 1.46-fold (95% CI 1.23-1.74) while lowering the hazard of low engraftment (HR 0.61, 95% CI 0.36-1.04); a cumulative AUC of 4,800 mgh/L maximised the probability of success at 82% 2. An exposure-response relationship of that steepness is not the signature of a drug that does not need monitoring. It is the signature of a drug whose monitoring has not been built.
| target | DNA - alkylation by the spontaneously generated (2S,3S)-1,2-epoxy-3,4-butanediol monoepoxide and (2S,3S)-1,2:3,4-diepoxybutane, formed non-enzymatically at physiological pH. No repair-pathway node exists in this graph as a target. |
|---|---|
| modality | small molecule; a bifunctional alkylsulfonate PRODRUG whose activation is pH-driven and non-enzymatic, unlike cyclophosphamide (CYP-dependent) and unlike busulfan (not a prodrug at all) |
| role in this disease | alternative to busulfan in irradiation-free conditioning; not used in Interfant-06, which predates its adoption |
| us approval | NDA 214759, GRAFAPEX, Medexus Pharma; original submission 2025-01-21; Type 1 - New Molecular Entity |
| us labelled indication | allogeneic HSCT preparative regimen in combination with fludarabine, AML and MDS, adult and paediatric patients 1 year of age and older |
| us labelled dose | 10 g/m2/day i.v. over 2 hours on days -4, -3 and -2 |
| infant all status | OFF LABEL on two counts - the indication is AML/MDS and the minimum age is 1 year; an infant under 12 months with KMT2A-rearranged ALL satisfies neither |
| proposed target auc mg h per L | 4,800 cumulative (Chiesa 2020, derived in a mainly non-malignant paediatric cohort and restricted by its authors to non-malignant disease) |
| exposure mortality slope | HR 1.46 (95% CI 1.23-1.74) per additional 1,000 mg*h/L |
| vs busulfan | no detected difference on OS, EFS, relapse, NRM, GVHD or GRFS in 308 children with ALL over 4 years old in FORUM; 1 fatal VOD in each arm |
| vocabularies searched | PubMed [tiab] on 2026-09-20: treosulfan OR treosulphan OR dihydroxybusulfan 558; Ovastat 8; Trecondi 1. Generic-name retrieval is adequate; the asymmetry in this drug is REGULATORY and was only visible in openFDA Drugs@FDA, not in PubMed. |
key ae
[ "myelosuppression", "mucositis", "skin toxicity, particularly in occluded areas and the nappy area in small children", "veno-occlusive disease (reported, at a rate comparable to busulfan in FORUM)", "seizures (less associated than with busulfan; anticonvulsant prophylaxis is not label-mandated)", "secondary malignancy (alkylator class)" ]
resistance
[ "glutathione S-transferase conjugation of the epoxide metabolites", "enhanced interstrand cross-link repair", "no KMT2A-rearrangement-specific treosulfan measurement exists" ]
trials
[ "NCT01949129" ]
- Off-label is not a paperwork problem in a baby. No approved label covering an infant under 12 months means no required paediatric investigation plan, no labelled dose, no post-marketing paediatric safety commitment, and no regulatory forcing function to generate the data that is missing.
- 'No TDM needed' may be a description of what has not been built rather than of what the drug does. A 1.46-fold mortality hazard per 1,000 mg*h/L is a steep slope to be running unmonitored on a fixed body-surface-area dose in a patient whose surface-area-to-weight ratio is the most extreme it will ever be.
- The comparison that makes treosulfan attractive was made inside the arm that lost. Busulfan and treosulfan being indistinguishable is a finding within FORUM's chemoconditioning arm, which was significantly inferior to TBI in children over 4.
- Skin toxicity in occluded areas is an infant-specific hazard that the adult literature under-reports, because adults are not in nappies. The alkylating metabolites are excreted and concentrate where skin is occluded and moist.
- A prodrug whose activation is pH-driven will convert differently wherever pH differs - in an acidotic, septic or dehydrated infant the delivered alkylating exposure is not the infused dose, and nothing in fixed BSA dosing accounts for that.
- Has treosulfan ever been used to condition an infant under 12 months with KMT2A-rearranged ALL, and with what result? FORUM's floor is 4 years and the US label's is 1 year, so the published evidence in this exact population may be zero.
- Does treosulfan need therapeutic drug monitoring in leukaemia? The steep exposure-mortality slope says the question is open; the only proposed target was derived in, and restricted by its authors to, non-malignant disease.
- Treosulfan's activation is non-enzymatic and pH-driven, so unlike busulfan it should not depend on hepatic conjugating capacity. Does that make its pharmacokinetics MORE predictable in an infant with immature hepatic enzyme expression, which would be a genuine infant-specific argument for it? Nobody appears to have asked.
- Is there a hepatic-toxicity advantage at all? FORUM found one fatal VOD in each arm, which is the only head-to-head comparison and is uninformative at n=2.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In the prespecified comparison within the FORUM trial, 180 children with ALL receiving busulfan/thiotepa/fludarabine and 128 receiving treosulfan/thiotepa/fludarabine had comparable 3-year overall survival (0.71 vs 0.72), event-free survival (0.60 vs 0.55), relapse (0.31 vs 0.36), non-relapse mortality (0.08 vs 0.09) and GVHD-free relapse-free survival (0.48 vs 0.45), with one fatal veno-occlusive death in each group.
Comparable outcomes after busulfan- or treosulfan-based conditioning for allo-HSCT in children with ALL: results of FORUM. · Kalwak K, Moser LM, Poetschger U, Bader P, Kleinschmidt K, Meisel R, Dalle JH, Yesilipek A, et al. · Blood Advances 2025What this evidence cannot show
Prespecified but NOT randomised: patients received busulfan or treosulfan by centre practice and by contraindication, so the comparison is observational within a randomised trial and open to confounding by indication - a patient with hepatic concern is more likely to receive treosulfan. 308 patients total, and the event-free survival intervals (0.53-0.67 vs 0.46-0.63) are wide enough to hide a clinically meaningful difference in either direction. Absence of a detected difference is not equivalence. All patients were aged over 4 years at HSCT, because that is FORUM's eligibility floor, so nothing here applies to infants.
Evidence that argues against this (1)
- The same FORUM trial's primary randomised result is that BOTH chemoconditioning backbones are inferior to TBI plus etoposide in children over 4: 2-year overall survival 0.91 (0.86-0.95) with TBI versus 0.75 (0.67-0.81) with chemoconditioning, relapse 0.12 versus 0.33, and the trial stopped early on its futility rule. Establishing that busulfan and treosulfan are indistinguishable from each other is a statement made inside the losing arm. Peters C et al. 2021
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
In a prospective trial of 87 children receiving treosulfan-fludarabine conditioning, each 1,000 mg*h/L increase in treosulfan AUC(0-inf) raised the mortality hazard ratio to 1.46 (95% CI 1.23-1.74) and lowered the hazard of low myeloid chimerism to 0.61 (95% CI 0.36-1.04); a cumulative AUC(0-inf) of 4,800 mg*h/L maximised the probability of success at 82%. The paper states that treosulfan is given off-label in paediatric allogeneic transplant.
Proposed Therapeutic Range of Treosulfan in Reduced Toxicity Pediatric Allogeneic Hematopoietic Stem Cell Transplant Conditioning: Results From a Prospective Trial. · Chiesa R, Standing JF, Winter R, Nademi Z, Chu J, Pinner D, Kloprogge F, McLellen S, et al. · Clinical Pharmacology and Therapeutics 2020What this evidence cannot show
87 patients with 11 deaths and 12 low-engraftment events, so the exposure-response curves rest on small numbers; the 4,800 mg*h/L target was derived by maximising a composite success probability in the same dataset that generated it, with no external validation. The paper's own concluding sentence restricts its dosing recommendation to NON-MALIGNANT disease ('Measuring PK at the first dose and individualizing the third dose may be required in nonmalignant disease'), so applying the target to a leukaemia transplant extends it beyond what the authors claim. The composition of the cohort by age is not given in the abstract and infants are not separately analysed.
Evidence that argues against this (1)
- The largest paediatric ALL dataset reports comparable outcomes for treosulfan without any therapeutic drug monitoring, and the standing review summarises the field as 'comparable outcomes to busulfan-based conditioning in paediatric ALL, without the need for TDM to date'. An exposure-response relationship found in a mainly non-malignant cohort is not by itself evidence that monitoring would change outcomes in leukaemia, and no trial has tested treosulfan TDM against fixed dosing. Ben Hassine K et al. 2021
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
Treosulfan was first approved in the United States on 2025-01-21 under NDA 214759 (GRAFAPEX, Medexus Pharma), classified as a Type 1 New Molecular Entity, indicated in combination with fludarabine as an allogeneic HSCT preparative regimen for adult and paediatric patients one year of age and older with AML or with MDS, at 10 g/m2/day intravenously on days -4, -3 and -2.
GRAFAPEX (treosulfan) for injection - US approval record and prescribing information (NDA 214759; SPL set id 2d06b03c-17aa-0492-e063-6294a90abd8f) · Medexus Pharma, Inc.; US Food and Drug Administration · openFDA Drugs@FDA and Drug Label APIs, api.fda.gov 2025What this evidence cannot show
Queried live on 2026-09-20 against api.fda.gov; the original submission date 2025-01-21 and the class 'Type 1 - New Molecular Entity' come from the Drugs@FDA submissions record, the indication and dosing text from the Structured Product Label effective 2025-01-31. A label states what is approved, not what is done: off-label conditioning use in paediatric ALL is routine in Europe and is not constrained by this document. The record says nothing about European approval status, which was not checked in this beat, so the 'three decades in Europe' framing rests on the clinical literature rather than on a European regulatory record retrieved here.
Evidence that argues against this (1)
- The narrowness of the US label is partly a reflection of which trial was submitted, not of what the drug can do. The registration dataset was in AML and MDS, so the ALL evidence - including the 128 treosulfan patients in FORUM - was not the basis of approval and its absence from the label is procedural rather than a judgement about ALL. Reading the label as a statement about treosulfan's suitability in infant ALL would be a scope error of exactly the kind this project's brief warns about for statutes and database facts. Kalwak K et al. 2025
- E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.review
TBI-based conditioning is regarded as the standard for children over 4 years of age with ALL; the dominant myeloablative effect in chemotherapy-based conditioning comes from the alkylating agents, most commonly busulfan or treosulfan; busulfan has wide pharmacokinetic variability and a narrow therapeutic window requiring therapeutic drug monitoring, while treosulfan-based conditioning gives comparable outcomes in paediatric ALL without the need for TDM to date.
Total Body Irradiation Forever? Optimising Chemotherapeutic Options for Irradiation-Free Conditioning for Paediatric Acute Lymphoblastic Leukaemia. · Ben Hassine K, Powys M, Svec P, Pozdechova M, Versluys B, Ansari M, Shaw PJ · Frontiers in Pediatrics 2021What this evidence cannot show
A narrative review, cited here as a map of the field and for its explicit statement of the age-4 boundary, not as primary evidence for any numerical claim. Its 'without the need for TDM to date' is a statement about current practice and an absence of demonstrated need, not a demonstration that monitoring is unhelpful; the same group's later FORUM pharmacokinetic work shows how much a monitoring analysis can change a picture. Published 2021, so it predates both the treosulfan US approval and the FORUM busulfan exposure analysis.
Evidence that argues against this (1)
- A prospective paediatric trial in the same period reports a steep treosulfan exposure-mortality relationship (HR 1.46 per 1,000 mg*h/L, 95% CI 1.23-1.74) and proposes a target AUC, which is difficult to reconcile with 'no need for TDM'. The reviewers' claim is about outcomes not yet having been shown to improve with monitoring, and the pharmacokinetic data argue that the question is open rather than settled. Chiesa R et al. 2020
the two alkylator backbones of irradiation-free conditioning; indistinguishable on every FORUM endpoint in 308 children over 4 years old
alternative conditioning backbone, off-label in infant ALL on both diagnosis and age
the US treosulfan label mandates fludarabine 30 mg/m2/day on days -6 to -2 as its partner
TREO/THIO/FLU; the combined skin toxicity of the two alkylators is worst in occluded skin in children under 2