Oncofusion

start / graph / target-cd19-car-t-infant-kmt2a-r

Targetstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q6b-therapeutics2

CD19 CAR-T in infants with KMT2A-rearranged ALL — it works, it is manufacturable, and it does not hold

also called tisagenlecleucel · Kymriah · CTL019 · CD19 CAR T cells · huCART19

CD19 CAR-T is a treatment that reprograms a child's own T cells to hunt leukemia. In babies it can be made and it works at first, but most of the babies with KMT2A-rearranged leukemia eventually relapse, often with a leukemia that has changed lineage and no longer carries the CD19 target.

The same thing, for a scientist

Tisagenlecleucel was approved on the ELIANA trial, which excluded children under 3 years. Real-world European data in 35 infused patients under 3 (76% KMT2A-rearranged, 66% relapsed after prior allogeneic HSCT) show 12-month OS 84%, EFS 69%, but stringent EFS of only 41% — the gap between EFS and stringent EFS, driven by B-cell recovery and loss of persistence, is the whole story. Manufacturing is feasible: 94% of 146 batches for patients under 3 met specification globally. The dominant failure mode in KMT2A-rearranged disease is not simply CD19 antigen loss but myeloid lineage switch, a genotype-specific escape that no CD19-directed agent can follow.

Go deeper
Efficacy in the actual population1 deeper

An international retrospective cohort across 15 hospitals in 10 European countries screened 38 patients under 3 years at screening between September 2018 and September 2021; 35 (92%) received an infusion 1.

  • 29/38 (76%) had KMT2A-rearranged ALL; 25 (66%) had relapsed after prior allogeneic HSCT; median 2 prior lines of non-HSCT therapy; 18% had received inotuzumab, 37% blinatumomab.
  • MRD-negative complete response in 24 of 28 (86%) patients with measurable disease.
  • At median follow-up 14 months: 12-month OS 84% (95% CI 64-93), EFS 69% (47-83), stringent EFS 41% (23-58).
  • Probability of ongoing B-cell aplasia at 12 months: 70% (46-84).
  • CRS in 21/35 (60%), grade ≥3 in 5 (14%); neurotoxicity in 9 (26%), none severe.

Read the three survival numbers together. OS 84% and EFS 69% look like success. Stringent EFS 41% — which counts loss of B-cell aplasia and other markers of CAR loss as events — says that fewer than half of these infants had a durable, intact response at one year. Toxicity, which is the thing everyone worried about in babies, was manageable. Durability was not.

Manufacturing is not the barrier

Leukapheresis in a 10 kg infant was the stated reason ELIANA excluded this age group. In commercial practice, of 146 tisagenlecleucel quality-control batches for patients under 3 years (median age 1.2 years, median weight 10.4 kg in the US), 137 (94%) were manufactured within specification across 16 countries, with a trend toward improving CD3 counts and manufactured dose over 2017-2021 2. Patients weighing ≤10 kg trended toward needing one or more additional collection days.

So the objection that CAR-T is not feasible in infants is empirically false. The objection that it does not last is empirically true.

The genotype-specific failure mode

In B-ALL generally, CD19-directed relapse means CD19 antigen loss or CAR loss. In KMT2A-rearranged disease it also means myeloid lineage switch: the blast abandons the B-lineage program entirely and reappears as AML. This is not a theoretical concern — a post-tisagenlecleucel ALL-to-AML lineage switch patient-derived xenograft model pairing exists and has been used to test alternative CAR targets 3. This graph covers the mechanism separately.

The practical consequence is a sequencing problem the field has not solved. An infant treated with blinatumomab frontline (now likely, given Interfant-21) who relapses has had CD19 pressure applied once already. CD19 CAR-T applied second is a second round of the same selection. If lineage switch is the escape, the answer is not a better CD19 product.

What is being tried instead. FLT3-directed CAR T cells and bispecific CD19×FLT3 CAR T cells show potent preclinical activity against KMT2A-rearranged ALL, including in the post-tisagenlecleucel lineage-switch PDX pairing, exploiting the wild-type FLT3 overexpression that characterises this genotype 3. Clinically, NCT04276870 (CHOP, phase 2, recruiting, n=133) is a murine CART19 study with an explicit cohort for infants with very high risk KMT2A B-ALL — currently the main open CAR-T trial specifically naming this population 4.

Facts and figures
modalityautologous 4-1BB CD19 CAR T cells (tisagenlecleucel and investigational humanised products)
statusTisagenlecleucel approved up to age 25 for relapsed/refractory B-ALL; children under 3 were excluded from the registration trial ELIANA and use in infants is real-world/off-trial
pediatric eligibletrue
open nowtrue
manufacturing
median_age_years1.200median_age_years: 1.2median_weight_kg10.4median_weight_kg: 10.4within_spec_pct94.0within_spec_pct: 94batches146.0batches: 146
target genes
[
  "CD19"
]
trials
[
  "NCT04276870"
]
key results
{
  "n_infused": 35,
  "n_screened": 38,
  "kmt2a_r_fraction": "29/38 (76%)",
  "prior_hsct_pct": 66,
  "mrd_neg_cr": "24/28 (86%)",
  "os_12m_pct": 84,
  "os_ci95": [
    64,
    93
  ],
  "efs_12m_pct": 69,
  "efs_ci95": [
    47,
    83
  ],
  "stringent_efs_12m_pct": 41,
  "stringent_efs_ci95": [
    23,
    58
  ],
  "b_cell_aplasia_12m_pct": 70,
  "bca_ci95": [
    46,
    84
  ]
}
manufacturing — the numbers behind the chart
{
  "batches": 146,
  "within_spec_pct": 94,
  "median_age_years": 1.2,
  "median_weight_kg": 10.4
}
key ae
[
  "CRS any grade 60%, grade >=3 14%",
  "neurotoxicity any grade 26%, none severe",
  "prolonged B-cell aplasia"
]
resistance
[
  "myeloid lineage switch (genotype-specific)",
  "CD19 antigen loss",
  "loss of CAR persistence (the dominant signal here: stringent EFS 41% vs EFS 69%)"
]
How this could bite us back
  • Lineage switch: the escape route unique to this genotype, which converts a CD19-targetable B-ALL into an AML that no CD19 product can follow.
  • Antigen exhaustion of the pipeline: CD19 is used by blinatumomab and by CAR-T; spending it twice in one child's course is now routine and no equally effective second antigen is established in infants.
  • Persistence failure: the gap between 69% EFS and 41% stringent EFS is the honest measure of how often the product stops working, and it is the parameter the field under-reports.
  • Essential-function toxicity: prolonged B-cell aplasia during the first two years of life removes B cells during the period of primary immunisation; long-term humoral immunity in these children is unstudied.
  • Delivery: leukapheresis is feasible at 10 kg but patients under 10 kg need additional collection days, and the bridging period during manufacture is when a fast-proliferating infant leukemia can become unsalvageable.
  • Late effects: CRS and neurotoxicity in an infant brain, plus lymphodepleting fludarabine, have no long-term neurocognitive dataset in this age group.
Still open
  1. What is the KMT2A-rearranged-specific outcome in this cohort? 76% of patients had the genotype and the subgroup result was not reported; it is the number the lab most needs and it exists in the authors' dataset.
  2. How many of the events in the 41% stringent EFS were lineage switch rather than CD19-negative or CD19-positive B-lineage relapse? This single breakdown determines whether the answer is a better CAR or a different antigen.
  3. Does prior blinatumomab (37% of this cohort) reduce CD19 CAR-T durability? With Interfant-21 putting blinatumomab frontline, nearly every future infant reaching CAR-T will be CD19-pretreated.
  4. Can a menin inhibitor be used as a bridge to CAR-T, or does menin-driven differentiation change CD19 density in a way that compromises the product? Nobody has measured CD19 surface density under menin inhibition in primary infant ALL.
Every source behind this entry (4)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In an international retrospective cohort of children under 3 years receiving tisagenlecleucel (35 infused of 38 screened, 76% KMT2A-rearranged, 66% post-allogeneic HSCT), MRD-negative complete response occurred in 24/28 (86%) with measurable disease, and at median 14 months follow-up 12-month overall survival was 84% (95% CI 64-93), event-free survival 69% (47-83) and stringent event-free survival 41% (23-58), with CRS in 60% (grade ≥3 in 14%) and any-grade neurotoxicity in 26%.

    PMID 36084658DOI
    What this evidence cannot show

    Retrospective, n=35 infused, heterogeneous prior therapy, median follow-up only 14 months, and outcomes are not reported separately for the KMT2A-rearranged subgroup despite it being 76% of the cohort. Real-world cohorts of a licensed product are subject to selection at every step: who was screened, who was successfully bridged, who was infused. Three of 38 screened patients never received an infusion and are not in the survival analysis.

    Evidence that argues against this (1)
    • The response numbers should be read against the escape route CD19 CAR-T itself selects in this genotype. In two KMT2A-rearranged B-ALL patients treated with CD19 CAR-T, relapse occurred with complete loss of the B-lymphoid phenotype and outgrowth of clonally related CD19-negative acute myeloid leukemia, reproduced in xenografts. An 86% MRD-negative CR in a KMT2A-rearranged cohort therefore measures the fraction of disease that is currently CD19-positive, not the fraction that can be eliminated; the gap between 69% EFS and 41% stringent EFS in this same cohort is the visible part of that. Gardner R et al. 2016
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Of 146 tisagenlecleucel quality-control batches manufactured for patients under 3 years (median age 1.2 years, median weight 10.4 kg at US sites), 137 (94%) were manufactured within specification across 16 countries, with patients weighing 10 kg or less trending toward requiring additional collection days.

    Leukapheresis and Tisagenlecleucel Manufacturing Outcomes in Patients Age <3 Years with Relapsed/Refractory Acute Lymphoblastic Leukemia · Fong D, Tiwari R, Acker C, Clough L, Willert J · Transplantation and Cellular Therapy 2023
    PMID 37311511DOI
    What this evidence cannot show

    This is a manufacturer-authored analysis of commercial manufacturing data, with the conflict of interest that implies. Batches, not patients, are the unit of analysis, and the denominator counts only patients who reached leukapheresis — children too unwell or too small to be apheresed never enter it. Clinical outcomes were explicitly not reported.

    Evidence that argues against this (1)
    • Manufacturing success is not the binding constraint, and the clinical durability in this exact age group is much weaker than a headline EFS suggests. In the international retrospective cohort of children under 3 screened for tisagenlecleucel (15 hospitals, 10 countries; 38 screened, 35 infused, 76% KMT2A-rearranged), 12-month overall survival was 84% and event-free survival 69% — but STRINGENT event-free survival, which counts subsequent therapy and transplant as events, was only 41% (95% CI 23-58) with 18 events. Three of 38 screened patients never received an infusion. A 94% in-specification manufacturing rate and a 41% stringent EFS describe two very different bottlenecks, and the second is the one that matters. Ghorashian S et al. 2022
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    FLT3-directed CAR T cells and bispecific CD19xFLT3 CAR T cells show potent in vitro and in vivo activity against KMT2A-rearranged infant ALL, including in a post-tisagenlecleucel ALL-to-AML lineage-switch patient-derived xenograft model pairing, exploiting wild-type FLT3 overexpression in KMT2A-rearranged ALL, with minimal cross-reactivity against normal haematopoietic and non-haematopoietic tissues in vitro.

    PMID 35950535DOI
    What this evidence cannot show

    Preclinical. FLT3 is expressed on normal haematopoietic progenitors and dendritic cell precursors, so 'minimal cross-reactivity in vitro' is a weaker claim than on-target/off-tumour safety in a patient; prolonged FLT3 CAR persistence could plausibly produce durable progenitor aplasia. No clinical FLT3 CAR-T data in infant ALL exist.

    Evidence that argues against this (1)
    • Swapping the antigen does not escape the clonal structure that defeated the first antigen. Whole-genome phylogenetic reconstruction in a child who received anti-CD19 cellular and antibody therapy after multiply relapsed ALL showed that the AML which followed DERIVED FROM THE INITIAL ALL, carrying distinct driver mutations that were already detectable before the lineage switch emerged; the authors describe extensive prior clonal diversification followed by selection. A lineage-switch PDX pairing used to showcase an FLT3-directed product is therefore a model of selecting a pre-existing subclone, and the same selection logic applies to the second antigen: the relevant question is whether an FLT3-negative or FLT3-low subclone is already present, which a cross-reactivity experiment does not answer. Coorens THH et al. 2023
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.trial_registry

    NCT04276870 is a recruiting phase 2 study of murine CART19 at the Children's Hospital of Philadelphia (n=133) with an explicit cohort for infants with very high risk KMT2A B-ALL, alongside hypodiploid/t(17;19) B-ALL and CNS relapse cohorts.

    NCT04276870
    What this evidence cannot show

    Registry record only; single site, opened March 2020, primary completion March 2028, no results posted. The 133 enrolment figure covers all three orphan-indication cohorts, so the infant KMT2A cohort is much smaller.

    Evidence that argues against this (1)
    • The registry cannot tell you how much infant KMT2A data this study will produce, and its endpoint is the one the real-world data show is optimistic. Raw ClinicalTrials.gov v2 API, retrieved 2026-09-19: NCT04276870 has been RECRUITING since 2020-03-12 with ESTIMATED enrolment 133 across four orphan cohorts, minimum age 0 years and maximum 29 years, a SINGLE primary outcome of event-free survival at one year, primary completion 2028-03-10, study completion 2037-03-10, and hasResults FALSE. No cohort-level enrolment target is published in the record, so the size of the infant very-high-risk KMT2A B-ALL cohort is unknown — six years of recruiting with no posted results and an eight-year-later completion date is the realistic timeline. And one-year EFS is precisely the measure that fell from 69% to 41% under a stringent definition in the under-3 real-world cohort. Children's Hospital of Philadelphia 2020
Connected to