Oncofusion

start / graph / disease-nutm1-r-b-all

Diseasestrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q100-nutm1-comparator

NUTM1-rearranged B-ALL - the KMT2A-wild-type arm of the same infant trials, with 4-year EFS of 100% against 36-37% next door

also called NUTM1-r B-ALL · NUTM1-rearranged B-cell precursor acute lymphoblastic leukaemia · B-ALL with NUTM1 rearrangement · NUTM1-r BCP-ALL

About one in five babies with leukaemia who do NOT have the usual KMT2A rearrangement instead have a NUTM1 rearrangement. It is the same disease in the same age group, but almost all of these children are cured by ordinary chemotherapy, while the KMT2A group is the hardest leukaemia in paediatrics to cure.

The same thing, for a scientist

A WHO-5th-edition and ICC-2022 recognised independent subtype of B-ALL defined by fusion of one of at least twelve 5' partners to almost the whole NUTM1 open reading frame [E1][E5]. Its importance to this project is positional: NUTM1-r cases **are** the largest named genetic group inside the KMT2A-wild-type arm of the Interfant cohorts - **35 of 161 evaluated KMT2A-wildtype Interfant infants, 21.7%** - so the comparison with KMT2A-rearranged infant ALL is not a cross-study analogy but the other half of the same trials [E1]. Incidence is ~3-5% of infant B-ALL and 0.28-0.86% of paediatric B-ALL across three population-representative cohorts [E1]. Outcome is the headline: **4-year EFS 100% (n=35) in uniformly treated Interfant infants against 74.0% (95% CI 65.1-81.0) in 126 NUTM1-wildtype/KMT2A-wildtype infants, p=0.001**; 4-year EFS 92.1% (82.0-100) and OS 100% in 32 children aged 1 or over; every case reached complete remission; MRD below 0.05% in 41 of 45 (91.1%) against **48% in KMT2A-rearranged infant ALL**; and no patient was transplanted in first remission [E1]. An independent 175-infant series gives NUTM1-r 2-year EFS **80% (52-100), n=9**, against **36% (21-59)** for KMT2A::AFF1 and **37% (23-60)** for other KMT2A partners [E2]. The mechanism is p300-driven hyperacetylation rather than DOT1L-driven H3K79 methylation: global H3K27ac, new open chromatin regions co-opting B-lineage and stemness programmes including NF-kB and posterior HoxA, and **global DNA hypomethylation irrespective of the 5' partner**, with chemosensitivity attributed to a dependence on active transcription [E3]. Origin is **prenatal** [E4]. **The favourable label is a subtype average and it has exceptions with names**: a ZNF618::NUTM1 child who relapsed at 11 months and failed four CAR-T products, transplant and blinatumomab [E5]; two paediatric cases with high-risk features [E6]; and two adult cases [E7] in a subtype the 2021 international study said had never been reported in adults.

Go deeper
Why this entry is a comparator and not a curiosity

Infant ALL is split in this project's own framing into KMT2A-rearranged and everything else. The 'everything else' is about 20% of infant ALL 8 and has been poorly characterised throughout. NUTM1-r is now its largest named genetic group: 21.7% of evaluated KMT2A-wildtype Interfant infants [E2 for a second cohort: 9 of 33, 27%].

That makes every KMT2A-versus-comparator statement in this graph a statement partly about this entry. In particular:

  • The infant cohorts' control arms contain this disease, so a 'KMT2A-wild-type infant' comparator group is a mixture whose best-outcome component is NUTM1-r and whose second-best is PAX5-r (2-year EFS 100%, n=10) 2.
  • Age is not the explanation for the KMT2A-rearranged prognosis. NUTM1-r infants were younger than their KMT2A-wildtype/NUTM1-wildtype counterparts - 54% under six months against 16% - and still did better 1. In KMT2A-rearranged disease, age under six months is itself a high-risk feature. The same demographic that predicts disaster in one genotype predicts nothing in the other.
  • Both arise prenatally, from a single fusion event in a fetal progenitor 4. A prenatal origin, an infant presentation and a B-progenitor cell of origin are therefore not what makes KMT2A-rearranged infant ALL refractory.
The partner spectrum, and the one number this beat's dispatch brief got wrong

At least twelve 5' partners are reported in B-ALL: ACIN1, BRD9, CUX1, ZNF618, AFF1, ATAD5, CHD4, RUNX1, SLC12A6, IKZF1, KAT6A 5, plus SIN3A in a 2026 infant case 9.

Among 71 cases with a known partner in the international series 1:

infants (n=39)children >=1 y (n=32)
ACIN117 (44%)7 (22%)
BRD910 (26%)-
CUX16 (15%)9 (28%)
ZNF61819 (28%)

A correction to this beat's own dispatch premise. The brief stated, from the 175-infant series, that "the most common fusion is BRD9::NUTM1" 2. That is true of those nine patients and does not generalise: in the 85-case international series ACIN1 is the commonest infant partner at 44% with BRD9 second at 26%, and an independent 2025 paper describing a new cell line calls ACIN1::NUTM1 "the most frequent fusion" in NUTM1-R infant ALL 10. The graph should carry ACIN1 as the modal infant partner, with BRD9 as the modal functionally characterised one.

And there are two biological groups, split by HOXA9 rather than by age 1. A HOXA9-positive group - ACIN1, CUX1, BRD9, AFF1 - prevalent in infants under nine months; and a HOXA9-negative group, of which ZNF618 is the type case (HOXA9 low or absent in 4 of 5 paediatric cases), emerging near one year old and making up almost half of paediatric cases. Both groups upregulate BMI1 on 10p12.31-12.2, possibly through EP300 binding to the NUTM1 moiety 1.

The clinically relevant consequence: the reported poor-outcome cases are concentrated in the HOXA9-negative group - the ZNF618 case 5 and the CUX1 adult 7 - which is also the group with essentially no mechanistic work behind it.

The exceptions, in detail, because a subtype average is not a prognosis

Before 2025, fewer than 20 ZNF618::NUTM1 patients had been reported and all had successful outcomes 5. Then one did not, and the course is worth stating in full because it is the counterweight to every '100%' in this entry.

A 3-year-old boy, karyotype normal at diagnosis, negative on a 41-fusion RT-PCR panel and on FISH for eight loci - i.e. the fusion was cryptic and was only found by whole-transcriptome sequencing of the relapse sample, then confirmed retrospectively in the diagnostic marrow. CR after one induction course and four consolidations, then relapse at 11 months with del(6)(q13q23) and no pathogenic mutation on a 300-gene panel. Autologous CD19 CAR-T: ineffective at day 15. Autologous CD19/CD22 dual CAR-T: no CR. Blinatumomab plus cytarabine plus allogeneic CD19/22 CAR-T: CR. Allogeneic HSCT from a 5/10 matched father; relapse 7 months later; re-induction and donor infusion gave another CR with grade IV GVHD, and chimerism analysis showed loss of recipient-specific HLA on 6p, rendering the leukaemia resistant to graft-versus-leukaemia. Treatment was discontinued. The fusion is ZNF618 exon 10 to NUTM1 exon 2, a predicted 1,456-residue protein preserving both NUT domains 5.

Three things follow and all three matter to this graph. First, a favourable subtype label at a population level tells an individual family very little when the subtype's n is in the tens. Second, HLA loss on 6p after haploidentical transplant is an immune-escape route this graph records elsewhere for KMT2A-rearranged disease, and it is not genotype-specific. Third, four separate CAR-T infusions failed to hold this disease, which is a caution against assuming a chemosensitive subtype is also immunotherapy-sensitive.

Alongside it: two novel paediatric cases presenting with high-risk features 6, and two adult cases, the second reported explicitly as "a recurring pattern of poor prognosis" in adults 7. The 2021 international study stated that NUTM1 fusions "have at least so far not been reported in adults" 1. That statement is now false, and its falsification is the ordinary kind - the literature moved. Recorded so that nobody re-quotes the 2021 sentence.

Diagnostics: this subtype is routinely missed, and that has a direction

NUTM1 fusions are frequently cryptic: undetected by standard karyotyping and by targeted fusion panels. Two of the founding infant cases came from cryptic rearrangements invisible to standard cytogenetics 8; the ZNF618 child was negative on a 41-fusion RT-PCR screen, on eight-locus FISH and on karyotype, and was only diagnosed from whole-transcriptome sequencing 5.

Partner-independent detection is therefore the recommendation: break-apart FISH, RNA sequencing, NUTM1 immunohistochemistry, and NUTM1 RNA expression - 3' exon expression and protein expression are both highly specific to fusion cases, because the gene is otherwise silent outside testis 18.

The direction of the bias is favourable-to-unfavourable, which is the dangerous direction. A subtype detectable only by RNA-seq is enriched, in any retrospective series, for patients at centres that ran RNA-seq - and cases diagnosed at relapse by RNA-seq, like the ZNF618 child, were not NUTM1-r cases at diagnosis in anyone's prospective cohort. So the published 100% and 92.1% figures are drawn from case sets assembled partly by whichever technology was available, and the honest reading is that the favourable direction is robust while the point estimates are not. Boer 2021 says as much in a different way: "since not all NUTM1 fusions are obvious from the karyotype, the estimated frequency could be higher" 1.

What this disease does to the menin-interface hypothesis

Nothing directly - and something adjacent that deserves stating precisely.

The HOXA9-positive NUTM1 partner group runs a posterior-HOXA programme with no KMT2A fusion, no CXXC domain, no menin-binding motif and no reported menin contact. In the international series' BLUEPRINT case, an ACIN1-NUTM1 patient showed HOXA9 and HOXA10 expression "comparable with levels observed in two cases with KMT2A rearrangement", with decreased CpG methylation and high H3K4me3 and H3K27ac over that locus, plus an alternating active/repressive pattern in which two short HOXA9-overlapping regions escape otherwise constitutive Polycomb targeting 1.

So HOXA9 activation in an infant B-ALL does not require the KMT2A-menin interface. That is a constraint on the necessity half of the interface argument, not on the sufficiency half, and it says nothing about whether menin inhibition works in KMT2A-rearranged disease - where it demonstrably does. What it does is remove one available inference: "HOXA9 is high, therefore the KMT2A-menin complex is doing it" is not valid in infant B-ALL.

Boer 2021 draws the obvious next question and leaves it open: "HOXA upregulation has recently been associated with sensitivity to inhibitors of the KMT2A-Menin complex in KMT2A-rearranged acute leukemia; it would be interesting to assess whether HOXA-upregulating NUTM1-rearranged ALL cells are similarly sensitive" 1. Five years on, nothing read for this entry answers it. It is a cheap experiment with a KOPN32 cell line now available 10, and either answer is informative: sensitivity would relocate menin-inhibitor activity from the interface to HOXA dependence, and resistance would be the cleanest available demonstration that the interface, not the HOXA output, is what the drug class exploits.

Facts and figures
classificationWHO 5th edition (Classification of Lymphoid Neoplasms) and International Consensus Classification 2022 both recognise B-ALL with NUTM1 rearrangement as an independent subtype (PMID 40115133).
metric4-year event-free survival
value100
unit% (NUTM1-rearranged, n=35 uniformly treated Interfant infants; no events, so no lower bound is reported)
comparator value74
comparator unit% (NUTM1-wildtype/KMT2A-wildtype Interfant infants, n=126, 95% CI 65.1-81.0)
p value0.001
populationinfants under 1 year with KMT2A-wildtype B-cell precursor ALL, Interfant-99 and Interfant-06
regioninternational (Ponte di Legno Childhood ALL Working Group; Interfant study groups)
year2021
sourcePMID 34211097
immunophenotypeof 32 NUTM1-r infants with data: 16 pre-B, 12 common, 3 pro-B, 1 biphenotypic acute leukaemia. All three pro-B cases carried BRD9-NUTM1 (PMID 34211097).
fusion partners noteEleven from the 2025 literature review in PMID 40115133, plus SIN3A from a 2026 case report (PMID 41878999). The spectrum is still growing, which is the argument for partner-independent diagnostics. NOTE A MEASURED RULE-80 TRAP: SLC12A6[tiab] AND NUTM1[tiab] returns 0 with a fully parsed querytranslation - a genuine zero - while SLC12A6 is a genuinely reported partner. The partner appears in a review's table and not in any title or abstract. A bounded absence on a partner symbol would have been correct as a statement about titles and abstracts and wrong as a statement about the literature.
shared downstreamBMI1 upregulation on 10p12.31-12.2 occurs in BOTH subgroups and may be mediated by EP300 binding to the NUTM1 moiety (PMID 34211097); confirmed as a direct fusion-driven effect in an ACIN1::NUTM1-inducible model alongside HOXA9, HOXA10 and SKIDA1 (PMID 40884014).
epigenetic readoutGlobal enhancement of H3K27 ACETYLATION and GLOBAL DNA HYPOMETHYLATION irrespective of the 5' fusion partner, with distinctive open chromatin regions co-opting B-lineage and stemness programmes including NF-kB and posterior HoxA genes (PMID 41460961). Locus-level corroboration from an ACIN1-NUTM1 BLUEPRINT case: decreased CpG methylation around HOXA9 and HOXA10 with high H3K4me3 and H3K27ac (chromatin state E11, active TSS), inside a region otherwise constitutively Polycomb-targeted (state E7) with an alternating H3K4me1/H3K4me3 versus H3K27me3 pattern in which two short HOXA9-overlapping regions escape repression (PMID 34211097). THE CONTRAST WITH THE INDEX DISEASE IS THE MARK AND THE WRITER, NOT THE OUTPUT: KMT2A-rearranged disease reaches HOXA9 through DOT1L-deposited H3K79me2/3 and a menin-anchored complex; NUTM1-r reaches it through p300-deposited acetylation and no menin interface. The DNA-hypomethylation half has no established mechanism and neither NUTM1 nor any reported partner carries a DNMT or TET module.
cell typeB-cell precursor, most often pre-B or common, with pro-B confined to BRD9-NUTM1 in the cases reported; the mouse model of BRD9-NUTM1 is prepro-B-like. The cell of transformation is a FETAL B-cell progenitor (PMID 39099338).
cell stateB-lymphoid-committed and simultaneously leukaemic-stem-cell-like, and dependent on active transcription - which is offered as the mechanistic explanation for chemosensitivity (PMID 41460961).
diagnosticsFrequently CRYPTIC. Missed by standard karyotyping and by targeted fusion panels: the founding infant cases came from cryptic rearrangements (PMID 31515871) and the reported poor-outcome ZNF618 case was negative on karyotype, on eight-locus FISH and on a 41-fusion RT-PCR screen, and was diagnosed only by whole-transcriptome sequencing (PMID 40115133). Partner-independent detection is recommended: break-apart FISH, RNA sequencing, NUTM1 immunohistochemistry, and NUTM1 RNA expression (3' exon expression and protein expression are highly specific to fusion cases because the gene is otherwise silent outside testis).
ascertainment caveatDetection depends on RNA-seq availability, so retrospective series are enriched for RNA-seq-capable centres, and cases found only at relapse were never counted as NUTM1-r at diagnosis in any prospective cohort. The published point estimates (100%, 92.1%, 80%) therefore carry an ascertainment component in the favourable direction. The frequency estimate is explicitly a floor: 'since not all NUTM1 fusions are obvious from the karyotype, the estimated frequency could be higher' (PMID 34211097).
menin relationshipNO menin interface, NO CXXC domain, NO MBM-like motif in NUTM1 or in any reported partner read for this entry. Menin-inhibitor sensitivity in this subtype is UNTESTED and was raised as an open question in 2021 (PMID 34211097); a KOPN32 cell line is now available to test it (PMID 40884014).
why it matters to this projectThis is the contrast case for infant B-ALL at matched age, matched lineage and matched cohort. It rules out prenatal origin, infancy, a fetal B-progenitor cell of origin and a posterior-HOXA programme as sufficient explanations for the KMT2A-rearranged prognosis, because all four are shared. What is NOT shared is the mark, the writer, the menin interface and the chemosensitivity, and the most interesting unanswered question in this comparison is which of those four does the prognostic work.
epidemiology
{
  "share_of_evaluated_kmt2a_wildtype_infants_pct": 21.7,
  "share_of_evaluated_kmt2a_wildtype_infants_n": "35 of 161 (PMID 34211097)",
  "second_cohort_share_of_kmt2a_wildtype_infants_pct": 27,
  "second_cohort_n": "9 of 33 (PMID 40000924)",
  "share_of_infant_b_all_pct": "approximately 3-5% (PMID 34211097 discussion)",
  "share_of_paediatric_b_all_pct": "0.28-0.86% across three population-representative cohorts; roughly 0.4-0.9% in the discussion (PMID 34211097)",
  "adults": "none reported as of 2021 (PMID 34211097); two adult cases published by 2024 (PMID 39585385)",
  "median_age_infants_months": 5.6,
  "age_range_infants_months": "0.43-11.0",
  "median_age_children_years": 4,
  "age_range_children_years": "1-15",
  "under_six_months_pct": "54% of NUTM1-r KMT2A-wildtype infants versus 16% of NUTM1-wildtype/KMT2A-wildtype (p<0.0001)",
  "wbc_over_50_pct": "45% of infants versus 13% of children (p=0.002)",
  "origin": "prenatal - fusion and IG/TR markers detectable in cord blood at birth in one case (PMID 39099338)"
}
outcome detail
{
  "efs_4y_infants_interfant_pct": 100,
  "efs_4y_infants_comparator_pct": 74,
  "os_4y_infants_pct": 100,
  "os_4y_infants_comparator_pct": 88,
  "efs_4y_children_pct": 92.1,
  "efs_4y_children_ci": "82.0-100",
  "os_4y_children_pct": 100,
  "efs_2y_second_cohort_pct": 80,
  "efs_2y_second_cohort_ci": "52-100",
  "efs_2y_second_cohort_n": 9,
  "kmt2a_aff1_efs_2y_pct": 36,
  "kmt2a_other_partners_efs_2y_pct": 37,
  "mrd_under_0_05_pct": 91.1,
  "mrd_under_0_05_n": "41 of 45",
  "mrd_under_0_05_kmt2a_r_comparator_pct": 48,
  "transplant_in_cr1": "none",
  "complete_remission_rate": "all cases",
  "deletions_ikzf1_pax5_etv6_cdkn2ab": "rare - at least one deleted in only 4 of 38 cases analysed"
}
fusion partners reported
[
  "ACIN1",
  "BRD9",
  "CUX1",
  "ZNF618",
  "AFF1",
  "ATAD5",
  "CHD4",
  "RUNX1",
  "SLC12A6",
  "IKZF1",
  "KAT6A",
  "SIN3A"
]
partner frequency infants
{
  "ACIN1": "17 of 39 (44%)",
  "BRD9": "10 of 39 (26%)",
  "CUX1": "6 of 39 (15%)",
  "denominator": "39 infant cases with a known partner (PMID 34211097)"
}
partner frequency children
{
  "CUX1": "9 of 32 (28%)",
  "ZNF618": "9 of 32 (28%)",
  "ACIN1": "7 of 32 (22%)",
  "denominator": "32 children aged 1 or over with a known partner (PMID 34211097)"
}
biological subgroups
[
  {
    "name": "HOXA9-positive",
    "partners": [
      "ACIN1",
      "CUX1",
      "BRD9",
      "AFF1"
    ],
    "age": "prevalent in infants under 9 months",
    "note": "HOXA9 expression confirmed for ACIN1, CUX1, AFF1 and additionally found for BRD9 (PMID 34211097). The partner is presumed to supply a DNA-binding module that addresses the HOXA cluster."
  },
  {
    "name": "HOXA9-negative",
    "partners": [
      "ZNF618"
    ],
    "age": "emerges near 1 year and rises to almost half of paediatric cases",
    "note": "HOXA9 low or absent in 4 of 5 paediatric ZNF618-NUTM1 cases; the single infant with this fusion was 11 months old. This is also where the reported poor-outcome cases sit."
  }
]
epigenetic marks
[
  "H3K27ac",
  "DNA methylation (global, direction: DOWN)",
  "H3K4me3",
  "H3K4me1",
  "H3K27me3"
]
How this could bite us back
  • The intervention this subtype most needs is LESS treatment, and de-escalation trials fail in a specific way: they are powered on an EFS that is already at or near 100%, so any reduction can only be shown not to be worse, and a non-inferiority margin at n=35 per arm is wide enough to hide real harm. A de-escalation arm that loses two children would not be statistically distinguishable from the standard arm.
  • The favourable label is itself a hazard. It is drawn from case sets assembled where RNA-seq was available, it has published exceptions in both the paediatric and adult literature, and it is now written into two classifications. A clinician de-escalating on the label rather than on MRD is acting on an average over 85 patients.
  • Every mechanistically obvious drug for this disease reduces transcription - BET inhibitors, CDK9 inhibitors, p300 HAT inhibitors - and the stated mechanism of its chemosensitivity is a dependence on active transcription. The prediction is antagonism with the chemotherapy that currently cures it, and it should be tested in KOPN32 before any such combination reaches a child.
  • CD19-directed immunotherapy is the natural salvage and there is one documented case in which it failed comprehensively: two autologous CAR-T products, one allogeneic, blinatumomab, transplant, and finally HLA loss on 6p abolishing graft-versus-leukaemia. The salvage pathway for the rare NUTM1-r failure is therefore not established, and the subtype's chemosensitivity may have produced complacency about it.
  • Diagnosis depends on a technology, not on a morphology. Any claim this graph makes about the size of the KMT2A-wild-type infant group, or about what the comparator arm of an infant trial contains, inherits the sequencing coverage of the contributing centres - and an undiagnosed NUTM1-r infant is counted as a good-outcome B-other case, which inflates the comparator and shrinks the apparent contrast.
Still open
  1. Are NUTM1-rearranged blasts sensitive to menin inhibition? Raised explicitly in 2021 and still unanswered; a HOXA9-positive NUTM1 line (KOPN32) now exists. Either answer settles whether menin inhibitors exploit the KMT2A interface or HOXA dependence, and that is the single most valuable experiment this comparison suggests.
  2. Which of the four non-shared features - the mark, the writer, the menin interface, the transcription dependence - actually explains the outcome difference? Age, lineage, cohort and HOXA output are all matched between the two diseases, so the answer is inside a short list, and nothing read here narrows it.
  3. Can NUTM1-r infants be de-escalated? Every case reached CR, 91% had MRD below 0.05%, none was transplanted in CR1, and 4-year EFS in the uniformly treated arm was 100%. Boer 2021 raises the ETV6-RUNX1 precedent for treatment reduction. The obstacle is that the subtype is defined by a test many centres do not run, so a de-escalation protocol needs universal RNA-seq or NUTM1 IHC first.
  4. Why is the HOXA9-negative group where the failures cluster, and is that real or is it the arithmetic of who gets written up? The ZNF618 relapse and the CUX1 adult are both HOXA9-negative-group cases from overlapping author groups, and the HOXA9-negative group is also the majority of paediatric cases, so it has more patients to fail.
  5. Is this subtype immunotherapy-sensitive? One patient failed four separate CAR-T infusions - autologous CD19, autologous CD19/CD22, and allogeneic CD19/22 twice - before HLA loss on 6p after haploidentical transplant. n=1, but a chemosensitive leukaemia that resists CD19-directed therapy would be a mechanistically interesting dissociation and it has not been looked at.
Every source behind this entry (10)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In 85 NUTM1-rearranged B-ALL cases from the Ponte di Legno Childhood ALL Working Group and Interfant-99/-06, NUTM1 rearrangement accounted for 35 of 161 evaluated KMT2A-wildtype Interfant infants (21.7%) with a frequency of 0.28-0.86% of paediatric B-ALL across three population-representative cohorts; 4-year EFS was 100% in the 35 uniformly treated infants against 74.0% (95% CI 65.1-81.0) in 126 NUTM1-wildtype/KMT2A-wildtype infants (p=0.001, adjusted p=0.0001), 4-year OS 100% versus 88.0% (p=0.04), and in 32 children aged 1 or over 4-year EFS was 92.1% (82.0-100) with OS 100%; all cases reached complete remission, 41/45 (91.1%) had MRD below 0.05% against 48% in KMT2A-rearranged infant ALL, none were transplanted in first remission, 54% of NUTM1-r KMT2A-wildtype infants were under six months against 16% of NUTM1-wildtype (p<0.0001), ten partners were found among 71 cases (infants: ACIN1 44%, BRD9 26%, CUX1 15%; children: CUX1 28%, ZNF618 28%, ACIN1 22%), and the HOXA9-positive group (ACIN1, CUX1, BRD9, AFF1) is distinguished from a HOXA9-negative group typified by ZNF618.

    PMID 34211097DOI
    What this evidence cannot show

    Read in full from the Europe PMC open-access full text (PMC8478641). PubMed indexes NO abstract for this letter, so every number here is invisible to an abstract-only workflow - which is why this paper was absent from a graph that already cited the 175-infant series twelve times. Retrospective. Only the 35 Interfant infants are uniformly treated; the Ponte di Legno arm spans 1986-2019 and many protocols. A 100% Kaplan-Meier estimate at n=35 with no events does NOT have a lower confidence bound of 100%, and the paper reports no interval for it. The 74.0% comparator is NUTM1-wildtype/KMT2A-wildtype infants and NOT KMT2A-rearranged infants; the only within-paper KMT2A-rearranged comparison is the 48% MRD figure, which is quoted from a different study. Median follow-up in the 11 Ponte di Legno infants was only 1.7 years. Outcome is NOT reported by fusion partner, so no partner-specific prognosis exists.

    Evidence that argues against this (1)
    • The paper's statement that NUTM1 fusions 'have at least so far not been reported in adults' has since been falsified by two adult case reports, the second of which describes a recurring pattern of poor prognosis in adults. Chen X et al. 2024
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In 175 infants aged 0-365 days with BCP-ALL diagnosed 2010-2023 at the Dmitry Rogachev centre, genotyped by karyotype, FISH and RNA-seq with Sanger validation, the KMT2A-wild-type group was 33 cases (17.7%) containing NUTM1-r (n=9, 27%) and PAX5-r (n=10, 30.3%); 2-year EFS was 80% (95% CI 52-100) for NUTM1-r and 100% for PAX5-r, against 36% (21-59) for KMT2A::AFF1 and 37% (23-60) for all other KMT2A partners pooled (log-rank p=0.9 between KMT2A partners); the most common NUTM1 fusion in that group was BRD9::NUTM1 with breakpoints at BRD9 exons 3, 8 and 14.

    Genetic Diversity in KMT2A-r and KMT2A-Wt Groups: Assessing the Prognostic Value of Markers in BCP-ALL Among Infants · Ilyasova K, Zerkalenkova E, Soldatkina O, et al. · International Journal of Laboratory Hematology 2025
    PMID 40000924DOI
    What this evidence cannot show

    The NUTM1-r arm is n=9 with a confidence interval reaching 100%, and the authors write that 'the small cohort size limit the statistical power of the analysis'. The 80% versus 36% contrast must not be made to carry more weight than n=9 supports. Single centre, one country, retrospective. This graph ALREADY records these figures in epi-moscow-infant-bcp-all-aff1-versus-other-partners, which cites the same paper for the KMT2A-partner null; this entry cites it for the KMT2A-wild-type side and does not restate the partner-null finding. THE 'MOST COMMON FUSION' CLAIM IS COHORT-SPECIFIC AND IS CONTRADICTED BY LARGER SERIES - see [E1] and [E10], where ACIN1 is the modal infant partner.

    Evidence that argues against this (1)
    • ACIN1, not BRD9, is the commonest NUTM1 partner in infants in the 85-case international series (44% versus 26%) and in an independent 2025 report that calls ACIN1::NUTM1 the most frequent fusion in NUTM1-R infant ALL. This cohort's modal partner is a property of nine patients. Boer JM et al. 2021
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    NUTM1-rearranged leukaemia is a discrete entity with a unique transcriptional and epigenetic landscape featuring global DNA hypomethylation irrespective of the 5' fusion partner; the fusions drive B-lymphoid lineage commitment while conferring leukaemic stem cell properties, establish an aberrant chromatin state of globally enhanced H3K27 acetylation with distinctive open chromatin regions co-opting B-lineage and stemness programmes including NF-kB and posterior HoxA genes, and BRD9-NUTM1 alone induces serially transplantable prepro-B-like leukaemia in vivo; in stark contrast to resistant KMT2A-rearranged leukaemias, NUTM1-r cells are profoundly chemosensitive, mechanistically linked to a dependence on active transcription.

    Comprehensive molecular and functional analysis of NUTM1-rearranged leukemia · Nishimura K, Isobe T, Shigehiro T, et al. · Blood 2026
    PMID 41460961DOI
    What this evidence cannot show

    Read from the published abstract, not the full text. The assay behind 'global DNA hypomethylation', the number of partners profiled that licenses 'irrespective of the 5-prime fusion partner', the definition of 'profound sensitivity to chemotherapy', and the design of the in-vivo model are all unread here. This paper's abstract is also the source of the phrase 'in stark contrast to resistant KMT2A-rearranged leukemias', which is the strongest published statement of this entry's central contrast and is nonetheless a comparison drawn in a discussion rather than a head-to-head experiment.

    Evidence that argues against this (1)
    • An editorial comment on this paper appeared in the same journal and has not been read for this entry; its existence is recorded rather than the primary paper being presented as unchallenged. Shiloh R et al. 2026
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    NUTM1-rearranged infant BCP-ALL arises prenatally: in an umbilical cord blood sample taken at birth from one infant later diagnosed with NUTM1-r BCP-ALL, the patient-specific CUX1::NUTM1 fusion gene and two IG/TR leukaemic markers were already present and detectable.

    Prenatal origin of NUTM1 gene rearrangement in infant B-cell precursor acute lymphoblastic leukaemia · Bardini M, Fazio G, Abascal LC, et al. · British Journal of Haematology 2024
    PMID 39099338DOI
    What this evidence cannot show

    One patient, one fusion (CUX1::NUTM1). The authors describe it as a unique opportunity, which is the honest framing: cord blood from a child who later develops a specific leukaemia is rarely available. A single positive backtracking result establishes that a prenatal origin is possible in this subtype; it does not establish that all cases arise prenatally, and a negative result would not have established the opposite because of sensitivity limits.

    Evidence that argues against this (1)
    • A commentary in the same journal frames single-patient backtracking results as a 'battle of addition', i.e. accumulating case-level evidence rather than a settled class-level fact. British Journal of Haematology 2024
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    B-ALL with NUTM1 rearrangement is an independent subtype in both the WHO 5th edition classification of lymphoid neoplasms and the International Consensus Classification of 2022, with a reported incidence of 0.28-0.86% of paediatric B-ALL and eleven reported 5' partners in B-ALL (ACIN1, BRD9, CUX1, ZNF618, AFF1, ATAD5, CHD4, RUNX1, SLC12A6, IKZF1, KAT6A); fewer than 20 ZNF618::NUTM1 patients had been reported and all had successful outcomes until a 3-year-old boy whose fusion was cryptic - negative on karyotype, on an eight-locus FISH panel and on a 41-fusion RT-PCR screen, and found only by whole-transcriptome sequencing of the relapse sample then confirmed in the diagnostic marrow - relapsed 11 months after CR with del(6)(q13q23) and no pathogenic mutation on a 300-gene panel, failed autologous CD19 CAR-T and autologous CD19/CD22 dual CAR-T, achieved CR with blinatumomab plus cytarabine plus allogeneic CD19/22 CAR-T, relapsed 7 months after a 5/10 matched haploidentical transplant with loss of recipient-specific HLA on 6p conferring resistance to graft-versus-leukaemia, and had treatment discontinued after grade IV GVHD; the fusion was ZNF618 exon 10 to NUTM1 exon 2, a predicted 1,456-residue protein preserving both NUT domains.

    PMID 40115133DOI
    What this evidence cannot show

    A single case report, read in full from the Europe PMC open-access full text (PMC11925428); PubMed indexes no abstract for it. n=1, and a case report selected for publication BECAUSE its outcome was atypical, so it cannot be used to estimate how often this happens - the same authors' companion adult case [E7] makes clear they were collecting the atypical end. The eleven-partner list is the authors' literature review, not a primary enumeration, and it omits SIN3A [E9]. The 'both NUT domains preserved' statement is a predicted protein structure from the fusion transcript, not a measured one; it is nonetheless consistent with Pfam PF12881 matching twice in Q86Y26.

    Evidence that argues against this (1)
    • The same paper states that fewer than 20 ZNF618::NUTM1 patients had been reported and that all of them had successful outcomes, so this case is explicitly the first exception rather than evidence of a rate. The subtype-level favourable finding in the 85-case series stands. Boer JM et al. 2021
  6. E6weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

    Two novel cases of NUTM1-rearranged B-cell acute lymphoblastic leukaemia have been reported as presenting with high-risk features.

    Two novel cases of NUTM1-rearranged B-cell acute lymphoblastic leukaemia presenting with high-risk features · McClure BJ, Pal M, Heatley SL, et al. · British Journal of Haematology 2022
    PMID 34897646DOI
    What this evidence cannot show

    TITLE-ONLY. This is a letter with NO abstract indexed in PubMed and no open-access full text available through Europe PMC, so the only content verified for this entry is the record itself - PMID, title, journal, year, DOI and author list, all read programmatically out of the efetch XML with the path anchored to PubmedData/ArticleIdList/ArticleId (rule 81). Which high-risk features, which partners, and what the outcomes were are all UNREAD, and this evidence item must not be cited for any of them. It is included only to record that a high-risk-presentation report exists.

    Evidence that argues against this (1)
    • High-risk presenting features are not the same as a poor outcome; in the 85-case series every NUTM1-r patient reached complete remission and none was transplanted in first remission, so a high-risk presentation in this subtype has not been shown to predict a bad end. Boer JM et al. 2021
  7. E7moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    A CUX1::NUTM1-positive B-ALL in a 25-year-old woman relapsed early with a poor clinical outcome; it is reported as the second adult case of NUTM1 fusion-positive B-ALL and as contributing to growing evidence that adult cases may run a more aggressive course than paediatric ones.

    PMID 39585385DOI
    What this evidence cannot show

    A single case report, and the second of two adults in the world literature, so 'a recurring pattern' rests on n=2. The overlap with the ZNF618 author group means these are not independent observers of the atypical end of this subtype. It is load-bearing for one narrow purpose only: it falsifies the 2021 statement that NUTM1 fusions had not been reported in adults.

    Evidence that argues against this (1)
    • With n=2 adults in total, a poor outcome in both is entirely compatible with the subtype being favourable and adults being rare; the alternative explanation - that adult B-ALL is worse than paediatric B-ALL across nearly every genotype - is not excluded and is not addressed by a case report. Charlab R et al. 2023
  8. E8strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The subset of infant ALL without KMT2A rearrangement represents about 20% of infant ALL cases and is poorly characterised; two chemotherapy-sensitive non-KMT2A-rearranged infant ALL cases were found to carry identical ACIN1-NUTM1 fusions derived from cryptic chromosomal rearrangements undetected by standard cytogenetics, joining exon 3 or 4 of ACIN1 to exon 3 of NUTM1 and retaining the ACIN1 SAP DNA-binding domain plus most of NUTM1; partner-independent screening by NUTM1 immunohistochemistry and NUTM1 RNA expression is proposed for clinical use.

    Cryptic recurrent ACIN1-NUTM1 fusions in non-KMT2A-rearranged infant acute lymphoblastic leukemia · Pincez T, Landry JR, Roussy M, et al. · Genes, Chromosomes and Cancer 2020
    PMID 31515871DOI
    What this evidence cannot show

    Two patients. The prediction that NUTM1 fusions are recurrent in infant ALL was, at the time, an inference from public genomic datasets and a literature review; it has since been confirmed at scale [E1], so the inference is retrospectively vindicated rather than independently established here. 'Chemotherapy-sensitive' in two cases is an observation, not a survival estimate.

    Evidence that argues against this (1)
    • The proposed partner-independent screens rest on NUTM1 being silent outside testis, which is what makes 3' exon and protein expression specific. That specificity has not been tested against the NUTM2 family paralogues, which IntAct and UniProt both list as separate NUT-family genes, and cross-reactivity of a NUT antibody with a NUTM2 product would be invisible to this design. UniProtKB 2026
  9. E9weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

    A SIN3A::NUTM1 fusion has been reported in infant B-ALL, with durable molecular remission achieved on standard acute lymphoblastic leukaemia chemotherapy.

    PMID 41878999DOI
    What this evidence cannot show

    TITLE-ONLY. A letter with no abstract indexed in PubMed and no open-access full text retrieved; PMID, title, journal, year, DOI and authors were read programmatically from the efetch record with an anchored ArticleId path (rule 81). The breakpoints, the immunophenotype, the follow-up duration and what 'durable' means are all UNREAD and must not be cited from this item. It is included to record a twelfth reported partner, which the 2025 eleven-partner list [E5] does not contain, and to note that the partner spectrum is still growing - which is itself the argument for partner-independent diagnostics.

    Evidence that argues against this (1)
    • A single favourable case adds nothing to the subtype-level prognosis and, being a case report, is subject to the opposite publication bias from the poor-outcome reports. Both directions of case-level reporting in this subtype are selected. Chen X et al. 2025
  10. E10strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    ACIN1::NUTM1 is the most frequent fusion in NUTM1-rearranged infant ALL; KOPN32, established from a relapsed infant ALL case, is identified as the only cell line carrying a NUTM1 fusion, has few copy-number alterations like KMT2A-rearranged lines, and together with an ACIN1::NUTM1-inducible model shows direct fusion-driven upregulation of HOXA9, HOXA10, SKIDA1 and BMI1; KOPN32 is highly sensitive to both daunorubicin and vincristine, agents central to infant ALL induction.

    PMID 40884014DOI
    What this evidence cannot show

    One cell line and one inducible model, so the HOXA9/HOXA10/SKIDA1/BMI1 result is a direct causal demonstration in a single genetic background. 'The only cell line with NUTM1-fusion' is a statement about available lines as of 2025 and is the reason this subtype has almost no pharmacology: eight standard agents were tested in one line. Drug sensitivity in a relapse-derived line is not a clinical prognosis, and the reported daunorubicin/vincristine sensitivity is consistent with, but does not demonstrate, the subtype's clinical chemosensitivity.

    Evidence that argues against this (1)
    • KOPN32 was established from a RELAPSED infant ALL case, i.e. from a patient in whom chemotherapy had already failed, which is an awkward foundation for a model of a chemosensitive subtype and should temper any read-across from its drug-response profile to the clinic. Tamai M et al. 2025
Connected to