Oncofusion

start / graph / disease-infant-all

Diseaseestablished confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.Q7-clinical

Infant acute lymphoblastic leukaemia (ALL diagnosed under 1 year of age)

also called infant ALL · ALL in infancy · infant lymphoblastic leukaemia

Infant ALL is leukaemia of the blood-forming system diagnosed in a baby under one year old, and in about three quarters of these babies the cause is a broken KMT2A gene. It behaves like a different disease from the leukaemia of older children: the babies arrive sicker, most of them relapse while still on treatment, and only about half are alive six years later.

The same thing, for a scientist

Infant ALL (<1 year at diagnosis) is a biologically distinct entity dominated by KMT2A rearrangement (~70-80% of cases). Presentation is characteristically aggressive: very high leucocyte counts (median 145-213 x 10^9/L in the COG AALL0631 KMT2A-r cohort), frequent organomegaly, and CNS involvement in roughly half of KMT2A-rearranged cases by cytology. Interfant-06, the largest prospective series (n=651), reported 6-year EFS 46.1% (SE 2.1) and OS 58.2% (SE 2.0), unchanged from Interfant-99, with KMT2A rearrangement the strongest prognostic factor, followed by age, WBC and prednisone response. Two thirds of relapses occur within 2 years and usually on therapy. The first intervention in decades to move outcomes was the addition of one post-induction course of blinatumomab (2-year DFS 81.6% vs 49.4% historical control).

Go deeper
Why infant ALL is not just ALL in a small patient1 deeper

Three things separate it from the ALL of a 4-year-old, and all three follow from the biology rather than from the size of the patient.

  1. The lesion. ~70-80% carry a KMT2A rearrangement, compared with ~5% of ALL in older children 16. In the CCCG-ALL-2015 infant cohort 71 of 134 (53.0%) were KMT2A-rearranged, and outcome split sharply on that single variable: 5-year OS 35.6% (95% CI 25.7-49.4) for KMT2A-rearranged versus 69.8% (95% CI 59.3-82.1) for KMT2A-germline infants 4.
  2. The immunophenotype. The blast is arrested earlier in B-cell development, giving a CD19-positive, CD10-negative pro-B pattern with frequent aberrant myeloid antigen co-expression; over 90% of KMT2A-germline ALL is CD10-positive by contrast 6. See concept-flow-cytometry-immunophenotype-kmt2a-r.
  3. The clinical tempo. About two thirds of KMT2A-rearranged infants who relapse do so within 2 years of diagnosis, usually while still receiving treatment, and that relapse is generally refractory and fatal 5. There is no comfortable window in which to escalate therapy after a relapse; the first attempt has to work.
Where the boundary at 12 months comes from, and why it is soft

The under-1-year definition is an operational trial boundary, not a biological cliff. Within it, outcome grades continuously with age: in COG AALL0631 the high-risk group (KMT2A-rearranged and under 90 days) had 5-year EFS 17 +/- 7% against 40 +/- 5% for the 90-365 day intermediate-risk group 3. Interfant-06 uses 6 months rather than 3 as its clinical cut-point 1. Children over 1 year with an 11q23/KMT2A rearrangement are a separate and better-behaved population.

The numbers a family is actually quoted
CohortnEndpointResult
Interfant-06, all infants6516-year EFS / OS46.1% (SE 2.1) / 58.2% (SE 2.0) 1
Interfant-06, West European / North American groups-6-year EFS / OS49.4% (SE 2.5) / 62.1% (SE 2.4) 1
Interfant-06, high-risk group-6-year EFS20.9% (SE 3.4) 1
COG AALL0631, KMT2A-r IR+HR1465-year EFS / OS34 +/- 4% / 41 +/- 4% 3
CCCG-ALL-2015 infants1345-year EFS / OS52.1% (95% CI 44.2-61.4) / 66.1% (95% CI 58.4-74.8) 4
Interfant-06 + one blinatumomab course302-year DFS / OS81.6% (95% CI 60.8-92.0) / 93.3% (95% CI 75.9-98.3) 2

The last row is the only line in this table that improved on the one above it, and it did so with an immunotherapy rather than with more chemotherapy. Interfant-06 tested exactly the opposite hypothesis - that early intensification with myeloid-style courses would help - and found no difference (6-year DFS 39.3% for ADE+MAE versus 36.8% for lymphoid course IB, log-rank p = 0.47) 1.

What this means for a menin-directed programme

Three design constraints fall directly out of the clinical picture. First, the therapeutic window is the first year of treatment, because that is when the relapses happen 5; an agent that only works in the relapsed setting arrives too late for most of these babies. Second, CNS penetration matters, because roughly half of KMT2A-rearranged infants have CNS involvement at diagnosis by cytology 3. Third, the bar is not survival alone: the babies who do survive carry the late effects of intensive chemotherapy delivered to an organism that is still developing, so a targeted agent's value includes what it lets you remove from the backbone.

Facts and figures
icd10C91.0
age definitiondiagnosis before 12 months of age
dominant lesionfusion-kmt2a-aff1 and other KMT2A rearrangements
immunophenotypeCD19+ CD10-negative pro-B, frequent myeloid antigen co-expression
relapse timingtwo thirds within 2 years of diagnosis, usually on therapy
presenting wbc median x10e9 per L
cccg_all_2015_all_infants62.7cccg_all_2015_all_infants: 62.7cog_aall0631_intermediate…145.0cog_aall0631_intermediate_risk: 145cog_aall0631_high_risk213.0cog_aall0631_high_risk: 213
survival
Interfant-06 + blinatumom…2.000Interfant-06 + blinatumomab pilot (n=30): 2COG AALL0631 KMT2A-r (n=1…5.000COG AALL0631 KMT2A-r (n=146): 5CCCG-ALL-2015 infants (n=…5.000CCCG-ALL-2015 infants (n=134): 5Interfant-06 (n=651)6.000Interfant-06 (n=651): 6Interfant-06 high-risk gr…6.000Interfant-06 high-risk group: 6
kmt2a rearranged fraction PCT
[
  70,
  80
]
presenting wbc median x10e9 per L — the numbers behind the chart
{
  "cog_aall0631_intermediate_risk": 145,
  "cog_aall0631_high_risk": 213,
  "cccg_all_2015_all_infants": 62.7
}
cns involvement at diagnosis PCT
{
  "cog_aall0631_intermediate_risk_cns2_or_cns3": 48,
  "cog_aall0631_high_risk_cns2_or_cns3": 83
}
survival — the numbers behind the chart
[
  {
    "cohort": "Interfant-06 (n=651)",
    "years": 6,
    "efs_pct": 46.1,
    "os_pct": 58.2,
    "precision": "SE 2.1 / SE 2.0"
  },
  {
    "cohort": "Interfant-06 high-risk group",
    "years": 6,
    "efs_pct": 20.9,
    "precision": "SE 3.4"
  },
  {
    "cohort": "COG AALL0631 KMT2A-r (n=146)",
    "years": 5,
    "efs_pct": 34,
    "os_pct": 41,
    "precision": "+/- 4%"
  },
  {
    "cohort": "CCCG-ALL-2015 infants (n=134)",
    "years": 5,
    "efs_pct": 52.1,
    "os_pct": 66.1,
    "ci_efs": [
      44.2,
      61.4
    ],
    "ci_os": [
      58.4,
      74.8
    ]
  },
  {
    "cohort": "Interfant-06 + blinatumomab pilot (n=30)",
    "years": 2,
    "dfs_pct": 81.6,
    "os_pct": 93.3,
    "ci_dfs": [
      60.8,
      92
    ],
    "ci_os": [
      75.9,
      98.3
    ]
  }
]
How this could bite us back
  • Two thirds of relapses happen on treatment, so any new agent has to work in combination from the start; a sequential 'add at relapse' strategy will miss most patients.
  • Roughly half of KMT2A-rearranged infants have CNS involvement at diagnosis, so a targeted agent with poor CNS penetration will leave a sanctuary site that reseeds the marrow.
  • CD19-directed therapy is now part of the backbone, and lineage switch to a CD19-negative myeloid phenotype is a documented escape route in KMT2A-rearranged disease; it was rare (2% of Interfant-06 relapses) before immunotherapy was used routinely.
  • The population is small and split across consortia with different risk definitions, so an underpowered trial can make an effective agent look inert.
Still open
  1. Interfant-06 showed that intensification does not help and the blinatumomab pilot suggests immunotherapy does; is the difference that chemotherapy cannot reach the KMT2A-rearranged clone's self-renewing compartment, or that it kills it only at a dose infants cannot survive?
  2. Why does outcome grade so steeply with age within the first year when the fusion is the same? Is the determinant the cell of origin, the residual fetal transcriptional programme, or the dose reduction that very young infants receive?
  3. In an era of blinatumomab and menin inhibition, is the 12-month trial boundary still the right way to define this population, or should it be defined by immunophenotype and MRD kinetics instead?
Every source behind this entry (6)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Interfant-06 enrolled 651 infants with ALL and reported 6-year EFS 46.1% (SE 2.1) and OS 58.2% (SE 2.0); risk groups were low risk (KMT2A-germline), high risk (KMT2A-rearranged, older than 6 months with WBC >= 300 x 10^9/L or poor prednisone response) and medium risk (all other KMT2A-rearranged); myeloid-style consolidation was not superior to lymphoid course IB (6-year DFS 39.3% vs 36.8%, p = 0.47); high-risk 6-year EFS was 20.9% (SE 3.4); KMT2A rearrangement was the strongest prognostic factor, followed by age, WBC count and prednisone response.

    PMID 31283407DOI
    What this evidence cannot show

    Survival is reported with standard errors rather than confidence intervals. The 18 participating groups differ in supportive care, and the paper itself notes a 10-12% survival gap between West European/North American groups and the rest, so the pooled figure is not transportable to any single health system. Only 46% of high-risk patients assigned to transplant actually received it because of early events, so the high-risk EFS conflates biology with failure to deliver the intended therapy.

    Evidence that argues against this (3)
    • A trial cohort is not the disease population. Interfant-06 enrolled infants who were diagnosed at participating centres, survived to enrolment and consented, so early deaths and non-enrolled infants are missing from the denominator, and EFS on a protocol is not EFS in the world. Its own randomisation was informative in the negative direction: the trial did not show a benefit for the intensified myeloid-style arm it was designed to test. Journal of Clinical Oncology 2019
    • These figures are also being superseded rather than merely qualified. Adding blinatumomab to the Interfant-06 backbone in infants with KMT2A-rearranged ALL produced substantially better outcomes than the historical protocol results quoted here, so a 6-year EFS of 46.1% describes the pre-immunotherapy era. Quoting it as the current prognosis for a newly diagnosed infant would be wrong. New England Journal of Medicine 2023
    • The comparison across Interfant generations is not clean either. Interfant-99 reported 4-year EFS of 47.0% on 482 infants, which is close to Interfant-06's 6-year 46.1% at a different time point, so the apparent stability across protocols partly reflects different follow-up horizons rather than equivalent outcomes. The Lancet 2007
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Adding one post-induction course of blinatumomab to the Interfant-06 backbone in 30 newly diagnosed infants with KMT2A-rearranged ALL produced no toxic effect meeting the primary endpoint, 28 of 30 (93%) were MRD-negative or had MRD < 5 x 10^-4 after the infusion, and 2-year DFS was 81.6% (95% CI 60.8-92.0) versus 49.4% (95% CI 42.5-56.0) for Interfant-06 historical controls, with OS 93.3% (95% CI 75.9-98.3) versus 65.8% (95% CI 58.9-71.8).

    Blinatumomab Added to Chemotherapy in Infant Lymphoblastic Leukemia · van der Sluis IM, de Lorenzo P, Kotecha RS, et al. · New England Journal of Medicine 2023
    PMID 37099340DOI
    What this evidence cannot show

    Single-arm, n = 30, median follow-up 26.3 months, compared against historical rather than randomised controls. The confidence interval on 2-year DFS runs from 60.8% to 92.0%, so the true effect could be considerably smaller than the point estimate. Historical-control comparisons in infant ALL are vulnerable to improvements in supportive care over the same interval. **Regraded moderate -> established on 2026-09-20** under `concept-the-confidence-grading-rule`: this is the trial's own reported result, which is established as a fact about what the trial found.

    Evidence that argues against this (1)
    • The Interfant-06 randomised comparison shows that a plausible-looking intensification can produce no benefit at all, which is the base rate this single-arm result should be judged against. source 2019
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In COG AALL0631, 146 KMT2A-rearranged infants had 5-year EFS 34 +/- 4% and OS 41 +/- 4%; the high-risk group (KMT2A-r, age < 90 days, n = 35) had EFS 17 +/- 7% and OS 20 +/- 7% versus 40 +/- 5% and 48 +/- 5% for the intermediate-risk group (KMT2A-r, age 90-365 days, n = 111); median presenting WBC was 145 x 10^9/L (range 2-4335) in the intermediate-risk and 213 x 10^9/L (range 12-1800) in the high-risk group; CNS status was CNS1 in 50% of intermediate-risk and 17% of high-risk infants, CNS2 in 34% and 57%, and CNS3 in 14% and 26%.

    PMID 33623141DOI
    What this evidence cannot show

    The trial's own question (adding lestaurtinib) was negative, so these are the characteristics and outcomes of a cohort receiving an intensive COG backbone, not of a successful intervention. Survival is given as point estimate +/- standard error. CNS2 (low-level blasts in CSF without pleocytosis) is a less certain category than CNS3 and its reported frequency is sensitive to cytospin technique and to traumatic taps in a small infant.

    Evidence that argues against this (1)
    • The age-90-day boundary is a chosen cut-point on a continuous variable and the high-risk group it defines has 35 patients, so EFS 17 +/- 7% carries an interval from roughly 3% to 31%. The contrast with the intermediate group is real; its magnitude, and the position of the threshold that produced it, are not established by this analysis. Statistics in Medicine 2006
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In the CCCG-ALL-2015 infant cohort (n = 134, median age 254 days), median WBC was 62.7 x 10^9/L (range 1.3-777) with 11.2% having WBC >= 300 x 10^9/L; 71 (53.0%) were KMT2A-rearranged with partners AFF1 53.3%, MLLT3 25%, MLLT1 16.7% and MLLT10 5%; 5-year EFS was 52.1% (95% CI 44.2-61.4) and OS 66.1% (95% CI 58.4-74.8) overall, with 5-year OS 35.6% (95% CI 25.7-49.4) for KMT2A-rearranged versus 69.8% (95% CI 59.3-82.1) for KMT2A-germline infants.

    PMID 38634141DOI
    What this evidence cannot show

    A single-consortium cohort in China; the 53% KMT2A-rearranged fraction is lower than the 70-80% usually quoted, which may reflect ascertainment, referral pattern or detection method rather than a real population difference, and it makes the pooled EFS look better than KMT2A-r-only series. Median WBC of 62.7 x 10^9/L is also much lower than the COG KMT2A-r medians, consistent with the cohort containing many KMT2A-germline infants.

    Evidence that argues against this (1)
    • CCCG-ALL-2015's better survival is not evidence of better treatment. Only 53% of its 134 infants are KMT2A-rearranged against 100% of the AALL0631 cohort, and a KMT2A-germline infant has a substantially better outcome, so most of the difference between 52.1% and 34% EFS is cohort composition. The partner distribution is also detection-limited by whatever assay the trial used. Leukemia 2021
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.review

    About 75% of infant ALL is driven by KMT2A rearrangement; roughly two thirds of KMT2A-rearranged infants who relapse do so within 2 years of diagnosis while still on active treatment, usually with refractory and fatal disease; lineage switch accounted for only 2% (5 of 231) of first relapses in Interfant-06; and 80% of KMT2A-rearranged infants still have detectable MRD at the end of induction, half of them at high levels.

    KMT2A-rearranged acute lymphoblastic leukemia in infants: current progress and challenges · Stutterheim J, van der Sluis IM, Pieters R, et al. · Haematologica 2025
    PMID 40207717DOI
    What this evidence cannot show

    A narrative review by the Interfant investigators, so it is a map of their own trial programme rather than an independent synthesis; percentages are summary statements whose primary sources must be read for the exact denominators. The 2% lineage-switch figure predates widespread CD19-directed therapy and the same review cautions it may rise.

    Evidence that argues against this (1)
    • A review aggregating four quantities from four different sources gives each the same apparent standing. The lineage-switch figure is the one most likely to be an underestimate: 5 of 231 first relapses in Interfant-06 predates routine CD19-directed therapy, and this graph's own entries record lineage switch as a recognised failure mode of blinatumomab and CD19 CAR-T in KMT2A-rearranged infants. A historical rate cannot be carried into the immunotherapy era. Leukemia 2021
  6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.review

    KMT2A rearrangement is present in up to 80% of infant ALL, with partner frequencies AFF1 48%, MLLT1 24%, MLLT3 16%, MLLT10 6% and other 6%; the blasts are CD19-positive and CD10-negative with frequent myeloid antigen co-expression, whereas over 90% of KMT2A-germline ALL is CD10-positive.

    Infant Acute Lymphoblastic Leukemia-New Therapeutic Opportunities · Kulczycka M, Derlatka K, Tasior J, et al. · International Journal of Molecular Sciences 2024
    PMID 38612531DOI
    What this evidence cannot show

    A review; the partner frequencies are re-quoted from the Meyer/Marschalek recurrent-rearrangement series and differ by a few percentage points between publications and between referral cohorts. The 'up to 80%' upper bound is the top of a range, not a central estimate.

    Evidence that argues against this (1)
    • The CCCG-ALL-2015 infant cohort found only 53.0% KMT2A-rearranged, well below the quoted 70-80%, which shows the fraction is cohort-dependent. source 2024
Connected to