Oncofusion

start / graph / risk-lineage-switch-cd19-directed-therapy

Riskstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q4-cells

Lineage switch as antigen escape from CD19-directed therapy in KMT2A-rearranged leukemia

also called post-immunotherapy lineage switch · CD19-negative myeloid escape · blinatumomab-induced lineage switch · CAR-T lineage switch · myeloid escape after CD19 targeting

Drugs that hunt the CD19 marker on B-cell leukemia work extremely well, but in KMT2A-rearranged leukemia the cancer can dodge them by turning into a myeloid leukemia that has no CD19 at all. It usually happens within weeks to a couple of months, it is often missed because the B-cell test comes back clean, and most patients do not survive it.

The same thing, for a scientist

Lineage switch (LS) is the defining non-CD19-loss mechanism of antigen escape from CD19-directed immunotherapy (blinatumomab, CD19 CAR T cells, CD19 antibody-drug conjugates) in KMT2A-rearranged (KMT2A-r) B-ALL. KMT2A rearrangement is the only pre-infusion feature reproducibly associated with LS-pattern relapse after CD19 CAR T cells; LS accounts for roughly 7% of post-CAR relapses overall but is heavily concentrated in KMT2A-r and infant disease, where reported incidence spans 0-28.6% across retrospective cohorts. LS emerges early — median 1 month after blinatumomab and 2.8 months after CAR T cells — frequently from a state of documented MRD-negative remission of the B-ALL, and carries remission rates under 40% and median overall survival of 4.8 months. Counterbalancing this, frontline blinatumomab added to Interfant-06 chemotherapy in 30 KMT2A-r infants produced a 4-year DFS of 83.3% versus 44.0% in controls with no lineage switch reported, so the risk is real but does not outweigh the benefit of CD19 targeting in this population.

Go deeper
The numbers, from cohorts with denominators3 deeper

Post-CAR relapse patterns (n = 420 treated, Pediatric Real-World / multicentre murine-based CD19 CAR cohort). Of 420 children and young adults with relapsed/refractory B-ALL treated with a murine-based CD19 CAR construct, 166 (39.5%) relapsed: 83 (50%) CD19-positive, 68 (41%) CD19-negative, and 12 (7.2%) lineage switch. Among all pre-infusion variables tested, the presence of a KMT2A rearrangement was the only pre-infusion risk factor associated with lineage switch. Median overall survival after any post-CAR relapse was 11.9 months (95% CI 9-17), and among patients who experienced a lineage switch there were no long-term survivors 1.

KMT2A-r infant ALL specifically (n = 231 CAR-treated, of whom 13 KMT2A-r infant ALL). In a hierarchical cytogenetic analysis of 231 children and young adults on five CD19-directed CAR trials or commercial tisagenlecleucel, overall complete remission was 94% with no difference between risk strata, and for the 13 patients with KMT2A-rearranged infant ALL 2-year relapse-free survival was 67% (95% CI 45-99) and overall survival 62% (95% CI 40-95). Multivariable analysis showed no increased risk of relapse (HR 0.70, 95% CI 0.21-2.90, P = .7040) but a higher proportion of relapses associated with myeloid lineage switch and a 3.6-fold increased risk of all-cause death (95% CI 1.04-12.75, P = .0434) 2.

That pairing is the key clinical message: CAR T cells get KMT2A-r infants into remission as reliably as anyone else, but when those patients do relapse, the relapse is more often a lineage switch and it kills them.

Incidence, honestly stated. Project EVOLVE states explicitly that the incidence of post-immunotherapy LS in patients with KMT2A-r or infant ALL 'is not fully elucidated, and has ranged from 0% to 28.6% after CD19-directed CAR T cells across retrospective patient cohorts' 3. A 0-28.6% range is the honest state of knowledge; anyone quoting a single incidence figure is over-reading the literature.

Timing: blinatumomab is faster than CAR T

Across the 70 primary-cohort EVOLVE cases, LS emerged at a median of 1.5 months (range 0-36.5) from the most proximal immunotherapy, with 81.4% within the first six months. Stratified by agent, the median time to LS was 1 month after blinatumomab (range 0-32, n = 31) and 2.8 months after CAR T cells (range 0.3-36.5, n = 34). At least seven cases emerged during the blinatumomab infusion itself, and at least six emerged before or at the one-month post-CAR time point. For the 13 cases that presented six months or more after immunotherapy, the median interval was 11 months (range 6.9-36.5) 3.

So the surveillance window that matters is measured in weeks, not years — and for blinatumomab it opens while the drug is still running.

It happens out of remission, which is why it gets missed

In the EVOLVE primary cohort, 38 of 70 (54.3%) patients had achieved an MRD-negative complete remission of their B-ALL after the most proximal pre-LS immunotherapy (response rates were higher after CAR T than after blinatumomab: 70.6% vs 35.5%, P = .0063) 3. In an independent 33-case pathology series, 79% (26/33) of patients were in complete remission of the first leukemia by flow cytometry at the moment the switch was identified 5.

This is the operational trap. A B-lineage-gated MRD panel run on a marrow that now contains monocytic blasts reports a clean result. Verifying the B-ALL is gone is not the same as verifying the marrow is clear.

Which agents, and whether CD19 is the specific culprit

In EVOLVE, the immunotherapy most proximal to LS was CAR T cells in 34 (48.6%), blinatumomab in 31 (44.3%), inotuzumab (CD22-targeted ADC) in 4 (5.7%) and an investigational CD19-targeted ADC in 1 (1.4%). Cumulative exposure before LS included blinatumomab in 45 (64.3%), CAR T cells in 35 (50%) and inotuzumab in 17 (24.3%); 11 (15.7%) had received both blinatumomab and CAR T cells. Of the CAR constructs in patients with LS, 29 (85.3%) were single-antigen CD19, 3 (8.8%) CD19/CD22 and 2 (5.9%) CD22 alone 3.

Two cautions follow. First, LS also occurred after CD22-directed therapy and, in the wider EVOLVE cohort, after CD38- and CD7-directed agents — so the driver is any single-lineage antigen pressure, not CD19 as a molecule. Second, the original 2016 observation was made in only seven patients: of seven MLL-rearranged B-ALL patients treated with CD19 CAR T cells, all achieved marrow complete remission by flow cytometry, and two developed AML clonally related to their B-ALL within one month of infusion 4. Two of seven is a small numerator on a small denominator, and it is the number most often cited as if it were an incidence rate. It is not.

Outcomes after the switch

Bad, consistently, across every series with follow-up.

  • Remission rates under 40% and median overall survival 4.8 months after LS diagnosis in the 70-patient EVOLVE cohort; the predominant cause of death was resistant disease 3.
  • Median survival 2.9 months after the switch in the 33-patient pathology series (12.3 months from the first leukemia) 5.
  • No long-term survivors among the 12 patients with LS-pattern relapse in the 420-patient CAR cohort 1.
  • All nine patients died in the pre-immunotherapy paediatric series, despite chemotherapy being switched to match the new lineage 6.

One therapeutic signal is worth flagging. In the 33-case series, among the eight patients who received allogeneic HSCT after the lineage switch, all seven with available information achieved complete remission, versus 1 of 19 who did not receive HSCT (P < 0.001), and HSCT was the sole independent prognostic factor on multivariable analysis (P < 0.001) 5. That is a retrospective comparison with severe selection bias — fit patients who achieve a response get transplanted — but it is the only intervention in this literature associated with survival.

Mitigation: what is actually supported2 deeper

1. Re-phenotype every lineage at every relapse and at every unexplained event, not just the B-cell compartment. This is the highest-yield, zero-cost intervention and every series recommends it. The specific triggers are: a marrow that looks 'in remission' by a B-gated panel while counts fail to recover; new monocytosis or monocytic blasts; any new neurological or focal symptom (see below). Zhou et al. conclude that the presence of risk factors 'warrants a thorough immunophenotypic evaluation of multiple cell lineages, particularly following therapy or at relapse' 5, and the EVOLVE consensus emphasises immunophenotypic and molecular monitoring because phenotype alone will miss 'lineage drift' 38.

2. Do not rely on a lineage-restricted MRD assay. In switch-prone BCP-ALL subtypes, B-cell-oriented flow cytometry demonstrably underestimated the MRD level relative to PCR, with discordance on day 15 for PAX5-P80R and on day 33 across DUX4r, PAX5-P80R and ZNF384r subtypes 7. Fusion-transcript or Ig/TCR PCR tracks the clone regardless of what it decides to look like.

3. Image and biopsy new focal symptoms; the switch can present extramedullary. In EVOLVE, CNS disease was present in 10 (14.3%) and non-CNS extramedullary disease in 5 (7.1%) at the time of the proximal immunotherapy 3. A reported case of KMT2A::AFF1 B-ALL on blinatumomab presented with back pain and lower-extremity weakness from an extradural thoracic mass causing spinal cord compression, which proved to be a mixed-phenotype myeloid sarcoma carrying the same KMT2A rearrangement, with concurrent MPAL in the marrow — the fourth reported myeloid sarcoma after blinatumomab 9.

4. Multi-antigen or myeloid-antigen-inclusive targeting — biologically motivated, not yet proven. EVOLVE recommends 'consideration of combined and/or multi-antigen approaches' while stating that this 'does not preclude use of antigen-targeted therapies in infant ALL' 3. Preclinically, FLT3 is constitutively overexpressed in KMT2A-r ALL, and bispecific CD19xFLT3 CAR T cells showed significant in vitro and in vivo activity against KMT2A-r ALL, with the monovalent FLT3 CAR active in — notably — a post-tisagenlecleucel ALL-to-AML lineage-switch patient-derived xenograft 10. That PDX pairing is the closest anyone has come to showing that a myeloid-inclusive target can retrieve a switched clone. It has not been tested in a patient.

5. Avoiding single-antigen pressure in KMT2A-r infants is NOT supported by the data, and this is where the received wisdom is wrong. See below.

The strongest counter-evidence: frontline blinatumomab in KMT2A-r infants

Thirty infants under one year with newly diagnosed KMT2A-rearranged ALL received Interfant-06 chemotherapy plus a single post-induction 28-day continuous infusion of blinatumomab (15 µg/m²/day). No toxic effect met the primary endpoint; 28 of 30 (93%) were MRD-negative or had MRD below 5x10⁻⁴ after the infusion; 2-year disease-free survival was 81.6% versus 49.4% in Interfant-06 historical controls, with overall survival 93.3% versus 65.8% 11.

With longer follow-up (median 4.2 years, range 3.2-6.0), 4-year DFS was 83.3% versus 44.0% in 214 selected Interfant-06 controls, 4-year OS 93.3% versus 60.2%, and the 4-year cumulative incidence of relapse 13.3% versus 49.6%. No further events occurred beyond the earlier report. No infection-related fatality occurred post-induction, against 4% in Interfant-06 12.

I read the full text of the long-term report: the words 'lineage', 'switch' and 'AML' do not appear anywhere in it 12. Thirty infants is far too small to exclude a several-percent lineage-switch risk, and absence of mention is not the same as systematic ascertainment. But the direction of the effect is unambiguous. 'Avoid single-antigen CD19 pressure in KMT2A-r infants' is not a defensible recommendation. The defensible version is: use it, and monitor all lineages aggressively while you do.

A second correction: lineage switch is not exclusive to KMT2A, and not always lethal

In 726 consecutively diagnosed children with BCP-ALL, 8% underwent a switch to the monocytic lineage with loss of CD19, and the genetic correlates were PAX5-P80R (5 of 5 cases), DUX4 rearrangement (30 of 41) and ZNF384 rearrangement (4 of 10) — not KMT2A. Despite substantial phenotypic change, and despite the challenge this posed to the appropriateness of BCP-ALL therapy, the monocytic switch was not associated with a higher incidence of relapse or a poorer prognosis on standard ALL treatment 7.

Two implications. First, a risk model keyed only to KMT2A will miss switch-prone patients. Second, the lethality attributed to 'lineage switch' may be partly the lethality of KMT2A-r relapsed disease under immunotherapy pressure rather than of lineage conversion itself.

Facts and figures
kindresistance
mechanismmechanism-lineage-switch-kmt2a
severityVery high. Remission rates <40% after switch; median overall survival 4.8 months (n=70, post-immunotherapy) and 2.9 months (n=33, all-comers); no long-term survivors among 12 lineage-switch relapses in a 420-patient CD19 CAR cohort. In KMT2A-rearranged infant ALL treated with CD19 CAR T cells, relapse risk is not increased but all-cause mortality is (HR 3.6, 95% CI 1.04-12.75).
detection trap54.3% of post-immunotherapy lineage-switch patients had achieved an MRD-negative complete remission of their B-ALL beforehand, and 79% of patients in an independent series were in flow-cytometric complete remission of the first leukemia at the moment of switch; B-lineage-gated flow cytometry underestimates MRD once a clone begins leaving the B lineage.
mitigation(1) Full multi-lineage immunophenotyping of every marrow at relapse, at unexplained cytopenia or failed count recovery, and at any new monocytosis — not a B-gated panel alone. (2) Track the clone with a lineage-independent molecular assay (KMT2A fusion transcript RT-PCR, Ig/TCR PCR or clonoSEQ) alongside flow. (3) Highest-vigilance window is 0-6 months after CD19-directed therapy, including during blinatumomab infusion itself. (4) Image and biopsy new focal or neurological symptoms; switch can present as extramedullary myeloid sarcoma. (5) Consider multi-antigen or myeloid-antigen-inclusive constructs (preclinical support for CD19xFLT3) rather than sequential single-antigen pressure. (6) Consider early allogeneic HSCT for those who do respond to switch-directed therapy. (7) Do NOT withhold CD19-directed therapy from KMT2A-rearranged infants on this basis: frontline blinatumomab plus Interfant-06 chemotherapy gave a 4-year DFS of 83.3% versus 44.0%.
monitoring window months0-6, with peak hazard 0-3
affects
[
  "drug-blinatumomab",
  "drug-tisagenlecleucel",
  "drug-inotuzumab-ozogamicin",
  "cd19-directed-car-t",
  "cd22-directed-therapy"
]
predisposing genotype
[
  "KMT2A rearrangement (any partner; t(4;11)/KMT2A::AFF1 most common)",
  "infant (<1 year) age at diagnosis",
  "BCR::ABL1",
  "DUX4 rearrangement",
  "EZH2 or RUNX1 mutation in B-ALL"
]
incidence
{
  "post_car_relapses_that_are_lineage_switch_pct": 7.2,
  "post_car_treated_patients_with_lineage_switch_pct": 2.9,
  "post_car_cohort_n": 420,
  "reported_range_in_kmt2a_r_or_infant_all_after_cd19_car_pct": "0 to 28.6",
  "kmt2a_r_share_of_post_immunotherapy_lineage_switch_pct": 64.3,
  "infant_share_of_post_immunotherapy_lineage_switch_pct": 28.6
}
time to event
{
  "median_months_overall": 1.5,
  "range_months_overall": "0 to 36.5",
  "median_months_after_blinatumomab": 1,
  "range_months_after_blinatumomab": "0 to 32",
  "n_blinatumomab": 31,
  "median_months_after_cd19_car_t": 2.8,
  "range_months_after_cd19_car_t": "0.3 to 36.5",
  "n_car_t": 34,
  "pct_within_6_months": 81.4,
  "cases_arising_during_blinatumomab_infusion": "at least 7"
}
salvage
{
  "first_line_lineage_switch_directed_cr_rate_pct": "<40",
  "allogeneic_hsct_after_switch": "7/7 informative transplanted patients achieved CR versus 1/19 non-transplanted (P<0.001); sole independent prognostic factor on multivariable analysis, but severely selection-confounded"
}
How this could bite us back
  • The resistance mechanism is invisible to the assay used to declare success: more than half of switch patients were MRD-negative for their B-ALL when the switch appeared. A protocol that escalates on MRD positivity will not see this coming.
  • Adding antigens may add selection surfaces rather than closing them. Lineage switch has been reported after CD22-, CD38- and CD7-directed agents too, so the vulnerability is single-lineage antigen pressure as a class, not CD19.
  • Myeloid-antigen-inclusive CARs inherit myeloid-lineage toxicity. FLT3 and most candidate myeloid antigens are expressed on normal haematopoietic stem and progenitor cells, so the construct that covers the escape route is the one most likely to cause prolonged marrow aplasia in an infant.
  • The switched cell may be immunosuppressive, so second-line T-cell-engaging salvage is being deployed into a marrow the leukemia has already turned against T cells.
  • Over-correcting is its own harm. The temptation to withhold CD19-directed therapy from KMT2A-rearranged infants would trade a 0-28.6% switch risk for the loss of a therapy that took 4-year DFS from 44.0% to 83.3% in exactly that population.
Still open
  1. What is the actual incidence of lineage switch in KMT2A-rearranged infants receiving frontline CD19-directed therapy? The published range is 0% to 28.6% and every number in it comes from a retrospective cohort with different ascertainment. The ongoing frontline blinatumomab trials could answer this at near-zero marginal cost by pre-specifying full multi-lineage immunophenotyping plus fusion-transcript PCR at every relapse and at every protocol marrow.
  2. Does bispecific or sequential myeloid-inclusive targeting actually prevent lineage switch, or merely shift the escape route? CD19xFLT3 CAR T cells are active preclinically including in a lineage-switch PDX, but a clone sitting at a multipotent progenitor stage may have more than two exits, and no trial has tested prevention as an endpoint.
  3. Is the switch preventable by not pushing the clone into deep remission with a T-cell-engaging agent while it is still bipotential? Switch is faster after blinatumomab (median 1 month, at least 7 cases during the infusion) than after CAR T cells (median 2.8 months) despite CAR T cells producing higher MRD-negative CR rates (70.6% vs 35.5%). Whether that reflects the kinetics of pressure, the depth of B-cell clearance, or the presence of persisting CAR T cells is unresolved and is testable by comparing switch rates against CAR persistence data.
  4. Should KMT2A-rearranged patients who achieve MRD-negative CR with CD19-directed therapy go straight to allogeneic HSCT? HSCT is the only intervention associated with survival after a switch has happened, but nobody has asked whether pre-emptive consolidative HSCT in the highest-risk genotype reduces switch-pattern relapse.
  5. Can the switch be detected before it is clinically apparent? EVOLVE describes 'lineage drift' — incomplete immunophenotypic change — as a separate category with equally poor outcomes. Whether drift is a reliable early warning that can trigger intervention, or simply an earlier-caught switch, has not been tested prospectively.
Every source behind this entry (12)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Among 420 children and young adults with relapsed/refractory B-ALL treated with a murine-based CD19 CAR construct, 166 (39.5%) relapsed — 83 (50%) CD19-positive, 68 (41%) CD19-negative and 12 (7.2%) lineage switch — with KMT2A rearrangement the only pre-infusion risk factor associated with lineage switch, a median overall survival after post-CAR relapse of 11.9 months, and no long-term survivors among those with lineage switch.

    Preinfusion factors impacting relapse immunophenotype following CD19 CAR T cells · Lamble AJ, Myers RM, Taraseviciute A, John S, Yates B, et al. · Blood Advances 2023
    PMID 35482927DOI
    What this evidence cannot show

    Multicentre retrospective review restricted to murine-based CD19 CAR constructs, so it does not describe humanised constructs or non-CAR CD19 agents. Twelve lineage-switch events is a small numerator, so 'KMT2A-r is the only preinfusion risk factor' is a statement about which associations survived testing in a small event count rather than proof that other factors do not matter. The 7.2% figure is a proportion of relapses, not a per-patient incidence: 12 of 420 treated patients is 2.9%.

    Evidence that argues against this (1)
    • Lineage switch in this setting need not be transdifferentiation of the dominant clone. Longitudinal sequencing across a post-immunotherapy B-ALL-to-AML switch showed shared ancestral alterations but distinct mutational profiles, with new TP53 and SMC1A mutations at switch and only a subset of relapse mutations retained, which the authors read as selective expansion of a genetically distinct subclone from a common progenitor. If that mechanism also operates in KMT2A-r disease, then 'switch' is partly a detection problem: the myeloid subclone was present and unmeasured before therapy. The case reported was not KMT2A-rearranged. Read: abstract only. Kondo K et al. 2026
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In 231 children and young adults on five CD19-directed CAR trials or commercial tisagenlecleucel, overall complete remission was 94% with no difference between cytogenetic risk strata; for the 13 patients with KMT2A-rearranged infant ALL 2-year RFS was 67% and OS 62%, with no increased risk of relapse (HR 0.70, 95% CI 0.21-2.90, P = .7040) but a higher proportion of relapses associated with myeloid lineage switch and a 3.6-fold increased risk of all-cause death (95% CI 1.04-12.75, P = .0434).

    PMID 34871373DOI
    What this evidence cannot show

    Thirteen KMT2A-r infant ALL patients: the 95% CIs on RFS (45-99%) and the hazard ratio for death (1.04-12.75) are enormous, and the mortality hazard barely clears significance. 37% of the full cohort was cytogenetically uninformative. Single-institution-led retrospective analysis across five trials spanning 2012-2019, so construct, manufacturing and supportive care varied. The 'higher proportion of relapses associated with myeloid lineage switch' is a qualitative statement in the abstract without a numerator I could verify.

    Evidence that argues against this (1)
    • Thirteen KMT2A-rearranged patients inside a 231-patient cohort. A 94% overall complete remission with no difference between cytogenetic strata is reassuring about response and uninformative about a rare event like lineage switch, for which 13 patients gives an upper bound near 20%. Journal of Clinical Oncology 2025
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In Project EVOLVE's 70-case primary cohort of post-immunotherapy lineage switch from B-ALL, 45 (64.3%) had KMT2A rearrangement, the most proximal immunotherapy was CAR T cells in 34 (48.6%) and blinatumomab in 31 (44.3%), median time to switch was 1.5 months overall (1 month after blinatumomab, 2.8 months after CAR T cells) with 81.4% within six months and at least seven cases arising during blinatumomab infusion, 38 (54.3%) had achieved an MRD-negative CR of their B-ALL beforehand, 20 (28.6%) were infants at initial diagnosis, remission rates after switch were under 40% and median overall survival was 4.8 months; the reported incidence of post-immunotherapy lineage switch in KMT2A-r or infant ALL has ranged from 0% to 28.6% after CD19-directed CAR T cells across retrospective cohorts.

    Project EVOLVE: an international analysis of postimmunotherapy lineage switch, an emergent form of relapse in leukemia · Silbert SK, Rankin AW, Hoang CN, Semchenkova A, Myers RM, et al. · Blood 2025
    PMID 40193715DOI
    What this evidence cannot show

    A solicited international case series with no denominator — it characterises lineage switch but cannot estimate its incidence, which the authors state explicitly. Ascertainment is biased toward centres with high-resolution flow cytometry and molecular monitoring and toward severe cases. Cases span 12 days to 76.5 years at initial diagnosis, pooling infant, paediatric and adult biology. The 0-28.6% incidence range is the authors' summary of the external literature, not a figure generated by this study.

    Evidence that argues against this (1)
    • At least one KMT2A-rearranged infant case switched to AML on blinatumomab and then spontaneously reverted to the lymphoid phenotype within weeks, without lineage-directed therapy. A cohort that ascertains switch at a single timepoint, as EVOLVE largely must, cannot distinguish a permanent lineage change from a transient, reversible phenotypic adaptation, and the reported incidence range of 0-28.6% is consistent with the endpoint itself being unstable. Read: abstract only. Wölfl M et al. 2018
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Of seven patients with MLL-rearranged B-ALL treated with CD19 CAR T cells, all achieved bone marrow complete remission by flow cytometry, but two developed acute myeloid leukemia clonally related to their B-ALL within one month of CAR T-cell infusion.

    PMID 26907630DOI
    What this evidence cannot show

    Seven patients. This is the index observation, not an incidence estimate, and the widely quoted '2 of 7' should never be presented as a rate. The cohort mixes infant and adult patients (MeSH terms include both Infant and Middle Aged) and reflects one centre's early CAR constructs and lymphodepletion regimen.

    Evidence that argues against this (1)
    • Clonal relatedness establishes shared ancestry, not the mechanism of the change. In murine ALL models under sustained CD19 CAR pressure, lineage switch was reproduced as reprogramming of the leukaemia — deletion of Pax5 or Ebf1 alone recapitulated it — and whether it occurred at all depended on the underlying oncogenic driver. Two clonally related leukaemias one month apart are compatible either with reprogramming or with outgrowth of a pre-existing myeloid-competent subclone, and the human data here do not separate them. Read: abstract only. Jacoby E et al. 2016
  5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In 33 clonally confirmed lineage switches, 79% (26/33) of patients were in complete remission of the first leukemia by flow cytometry at the time of the switch, median survival after switch was 2.9 months, and among the eight patients who received allogeneic HSCT after the switch all seven with available information achieved complete remission versus 1 of 19 who did not (P < 0.001), with HSCT the sole independent prognostic factor on multivariable analysis.

    Genetics and pathologic landscape of lineage switch of acute leukemia during therapy · Zhou T, Curry CV, Khanlari M, Shi M, Cui W, et al. · Blood Cancer Journal 2024
    PMID 38272888DOI
    What this evidence cannot show

    The HSCT result is severely confounded by selection: only patients well enough and responsive enough to proceed to transplant were transplanted, so the comparison 7/7 versus 1/19 is not a treatment-effect estimate. Retrospective case collection 2003-2022 across 19 institutions, median age 34.6 years, so only a minority are paediatric or KMT2A-r infant cases. Only 15 of 33 patients received any antigen-targeted immunotherapy, so this cohort is largely a chemotherapy-era description.

    Evidence that argues against this (1)
    • A genetically distinct subclone can also produce a clonally confirmed switch. Sequencing across one post-inotuzumab switch found shared ancestral alterations with a distinct mutational profile at switch, arguing for selection of a minor pre-existing subclone rather than direct transdifferentiation. 'Clonally confirmed' in these series generally means shared ancestry, which does not adjudicate between the two mechanisms. Read: abstract only. Kondo K et al. 2026
  6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In the pre-immunotherapy era, nine of 1,482 consecutive paediatric acute leukemias (0.6%) underwent lineage conversion, eight of nine in infants and seven of nine with 11q23/MLL abnormalities, confirmed at a median of 15 days from starting therapy, and all nine patients died despite chemotherapy being changed to match the new lineage.

    Lineage switch in childhood acute leukemia: an unusual event with poor outcome · Rossi JG, Bernasconi AR, Alonso CN, Rubio PL, Gallego MS, et al. · American Journal of Hematology 2012
    PMID 22685031DOI
    What this evidence cannot show

    Single-institution series in a middle-income setting; outcomes reflect that era's salvage options, and the same group reported no benefit of Interfant protocols over BFM-based protocols in their infant population, so 'all nine died' should not be read as the ceiling of what is achievable. It is cited here to establish that lineage switch pre-dates immunotherapy and therefore that CD19 targeting is an accelerant, not the sole cause.

    Evidence that argues against this (1)
    • Nine events in 1,482 leukaemias, seven with 11q23 abnormalities, detected at a median of 15 days. A pre-immunotherapy rate of 0.6% is the right comparator to have and it is built on nine cases, so it carries an interval wide enough to overlap several of the post-immunotherapy estimates this entry contrasts it with. Statistics in Medicine 2006
  7. E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Among 726 consecutive children with BCP-ALL, 8% underwent a switch to the monocytic lineage with loss of the B-cell immunophenotype including CD19, correlating with PAX5-P80R (5/5), DUX4 rearrangement (30/41) and ZNF384 rearrangement (4/10) rather than KMT2A; B-cell-oriented flow cytometry underestimated MRD in these patients, with PCR/flow discordance on day 15 for PAX5-P80R and on day 33 across all three subtypes, and the switch was not associated with higher relapse incidence or poorer prognosis on standard ALL therapy.

    DUX4r, ZNF384r and PAX5-P80R mutated B-cell precursor acute lymphoblastic leukemia frequently undergo monocytic switch · Novakova M, Zaliova M, Fiser K, Vakrmanova B, Slamova L, et al. · Haematologica 2021
    PMID 32646889DOI
    What this evidence cannot show

    This cohort is chemotherapy-treated, so it cannot test whether these same subtypes would switch lethally under CD19-directed antigen pressure — the prognostic neutrality may not transfer to the immunotherapy setting. Early monocytic switch detected by flow cytometry may be a milder, distinct entity from frank conversion to overt AML. The genetic correlations rest on small subgroup numbers (5 PAX5-P80R, 10 ZNF384r).

    Evidence that argues against this (1)
    • The study's own correlations are with PAX5-P80R, DUX4 and ZNF384 rather than with KMT2A, which makes it evidence that monocytic switching has drivers other than the one this entry is about — a useful control rather than support. Oncofusion knowledge graph (internal analysis) 2026
  8. E8moderate confidence 3 of 5: One strong study, or several smaller consistent ones.review

    An international expert group reviewing the biology and detection of lineage switch concludes that increasing use of CAR T cells and bispecific T-cell engagers, which apply selective antigen pressure, may increase the incidence of lineage switch across leukemia subtypes, that patients with lineage switch have dismal outcomes with unknown optimal therapy, and issues definitions and recommendations for immunophenotypic and molecular monitoring.

    Plasticity under pressure: biology and detection of lineage switch in acute leukemia · Kovach AE, Ding YY, Lamble AJ, Bernt K, Bomken S, et al. · Leukemia 2026
    PMID 42362807DOI
    What this evidence cannot show

    A review and expert consensus, included here only as a map of the recommended monitoring framework and not as primary evidence for any number. The claim that immunotherapy 'may increase the incidence' of lineage switch is explicitly hypothetical in the authors' own wording; no prospective study has established a rising incidence. I verified only the abstract, so the specific monitoring recommendations should be read from the paper before being operationalised.

    Evidence that argues against this (1)
    • The expert-group expectation that T-cell engagers will increase switch incidence sits awkwardly beside the possibility that some switches are transient. A KMT2A-rearranged infant who switched to AML during blinatumomab reverted spontaneously to ALL over weeks, so incidence estimates depend heavily on when and how often the phenotype is sampled, and 'dismal outcome' may partly select for cases that were ascertained because they persisted. Read: abstract only. Wölfl M et al. 2018
  9. E9weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

    A patient with refractory KMT2A/AFF1-rearranged B-ALL on the fourth cycle of blinatumomab developed back pain and lower-extremity weakness from an extradural thoracic mass causing spinal cord compression, which was a mixed-phenotype myeloid sarcoma carrying the same KMT2A rearrangement, with concurrent mixed-phenotype acute leukemia in the marrow; this was the fourth reported case of myeloid sarcoma following blinatumomab.

    PMID 41333696DOI
    What this evidence cannot show

    A single case report in a 19-year-old (not an infant), lost to follow-up, so no outcome data. It is cited only to document that post-blinatumomab lineage switch can present as extramedullary myeloid sarcoma with focal neurological signs, which is a surveillance point; it supports no frequency estimate. 'Fourth reported case' is the authors' own literature count and I did not independently verify the other three.

    Evidence that argues against this (1)
    • One patient. The case is well documented and establishes that myeloid sarcoma can emerge on blinatumomab with the rearrangement retained; it cannot speak to frequency, to predisposing features or to whether the site is preferential. Blood Cells, Molecules and Diseases 2010
  10. E10moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    FLT3 is overexpressed in KMT2A-rearranged ALL; monovalent FLT3-directed CAR T cells showed potent in vitro cytotoxicity and in vivo activity in xenograft models of KMT2A-rearranged infant ALL including a post-tisagenlecleucel ALL-to-AML lineage-switch patient-derived xenograft pairing, and bispecific CD19xFLT3 CAR T cells showed significant in vitro and in vivo activity against KMT2A-rearranged ALL, with minimal cross-reactivity against normal haematopoietic and non-haematopoietic tissues.

    PMID 35950535DOI
    What this evidence cannot show

    Entirely preclinical: cell lines and xenografts, no patient has received CD19xFLT3 CAR T cells for lineage switch. FLT3 is expressed on normal haematopoietic stem and progenitor cells, so 'minimal cross-reactivity' in murine xenografts is a weak predictor of human myelosuppression. The lineage-switch PDX is a single model. Whether a bispecific construct actually prevents switching, as opposed to treating it after the fact, has not been tested.

    Evidence that argues against this (1)
    • Preclinical, and the on-target problem is structural: FLT3 is expressed on normal haematopoietic stem and progenitor cells, so a FLT3-directed CAR attacks the compartment that must reconstitute the patient. Unlike B-cell aplasia after CD19 CAR-T, that toxicity has no supportive-care equivalent, and the randomised trial of FLT3 inhibition in this population was negative. Leukemia 2021
  11. E11established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Thirty infants under one year with newly diagnosed KMT2A-rearranged ALL given Interfant-06 chemotherapy plus one post-induction 28-day continuous blinatumomab infusion (15 microgram/m2/day) had no toxic effects meeting the primary endpoint, 28 of 30 (93%) were MRD-negative or had MRD below 5x10-4 after the infusion, and 2-year disease-free survival was 81.6% (95% CI 60.8-92.0) versus 49.4% (95% CI 42.5-56.0) in Interfant-06 historical controls, with overall survival 93.3% versus 65.8%.

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

    Single-arm, n = 30, compared with historical Interfant-06 controls rather than a randomised comparator, so the effect size is vulnerable to selection and era effects. Median follow-up at publication was 26.3 months. Thirty patients cannot exclude a lineage-switch risk of several percent, and the trial's primary endpoint was toxicity, not lineage surveillance. Confirmation in a randomised frontline trial is pending. **Regraded strong -> 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)
    • A negative result for switch in 30 infants does not mean blinatumomab cannot cause it in this disease. A KMT2A-rearranged infant treated with blinatumomab underwent an ALL-to-AML lineage switch that then spontaneously reverted; with n = 30 and an event that may be transient, the Interfant-06 blinatumomab cohort is underpowered to exclude a switch rate of a few per cent. Read: abstract only. Wölfl M et al. 2018
  12. E12strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    With a median follow-up of 4.2 years (range 3.2-6.0), the 30 blinatumomab-treated KMT2A-rearranged infants had a 4-year disease-free survival of 83.3% versus 44.0% in 214 selected Interfant-06 controls, 4-year overall survival of 93.3% versus 60.2%, and a 4-year cumulative incidence of relapse of 13.3% versus 49.6%, with no further events beyond the earlier report and no post-induction infection-related fatality against 4% in Interfant-06; the words 'lineage', 'switch' and 'AML' do not appear anywhere in the report.

    Sustained Benefit of Blinatumomab in Infants With KMT2A-Rearranged ALL: Long-Term Outcomes, Toxicity, and Pharmacokinetics · Vieira Martins M, de Lorenzo P, Kotecha RS, Attarbaschi A, Escherich G, et al. · Journal of Clinical Oncology 2026
    PMID 41505662DOI
    What this evidence cannot show

    The absence of any mention of lineage switch was confirmed by full-text search of the open-access version, but absence of reporting is not absence of events: the study was not designed to ascertain lineage switch, relapses are reported as relapses without immunophenotype detail in the main text, and n = 30 has no power to detect a low-single-digit switch rate. Comparison is against selected and matched historical Interfant-06 controls, not a randomised arm.

    Evidence that argues against this (1)
    • The absence of the words 'lineage', 'switch' and 'AML' from a report is weak evidence of absence. Switch after blinatumomab in KMT2A-rearranged infant ALL is documented, including a case that reverted spontaneously within weeks and so would be easy to miss or to record as a transient aberrant immunophenotype rather than as an event. Read: abstract only. Wölfl M et al. 2018
Connected to