Oncofusion

start / graph / hyp-ng2-directed-therapy-would-spare-the-candidate-stem-cell-fraction-in-t-4-11-leukaemia

Hypothesismoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q138-kmt2a-aff1-surfaceome

NG2 is absent from the candidate leukaemia stem-cell populations in t(4;11) ALL and 11q23 AML, so an NG2-directed therapy would debulk the disease and leave the fraction that initiates relapse

also called the NG2 stem-cell reservoir

NG2 marks the bulk of this leukaemia but not the small population of cells thought to restart the disease after treatment. A therapy aimed at NG2 would therefore clear the visible disease and leave behind the cells that cause relapse.

The same thing, for a scientist

In six AML/abn(11q23) and **three ALL/t(4;11)** samples analysed by four-colour flow cytometry, the candidate leukaemic stem-cell populations - CD34+CD38- in AML and **CD34+CD19-CD117+ in ALL** - **lacked NG2 expression in all patients**, while NG2 increased with the myeloid maturation marker CD33 on maturing cells [E1]. That is the strongest single argument against NG2 monotherapy and it is from 2006. It also sits against a 2018 result in the same disease: serial xenotransplantation of primary MLL-AF4-positive leukaemias showed NG2 is a **malleable marker that does not enrich for leukaemia-initiating or CNS-initiating cells** in infant MLL-rearranged B-ALL [E2]. The two agree that NG2 does not mark the initiating fraction, and the second adds that NG2 status is **not stable** - a cell can be NG2-negative and become NG2-positive. **Malleability cuts both ways**: it undermines NG2 as a stem-cell marker and it means the NG2-negative reservoir may not be a fixed, pre-existing clone but a state the disease re-enters. The patient measurement in this beat is consistent with a non-uniform antigen: NG2 is bimodal across KMT2A-rearranged TARGET samples, with one of four RNA-seq-profiled MLL-positive patients at 0.01 RPKM [E3].

Go deeper
What follows for trial design if this is right
  1. NG2-directed therapy should be assessed on depth of response and time to relapse, not on response rate. Debulking a bulk antigen looks excellent on day-29 marrow blast percentage and may not change event-free survival.
  2. It should be paired, with an agent covering the CD34+CD19-CD117+ fraction - which in this disease is where menin inhibition and the chromatin-directed agents this project studies would act.
  3. MRD by NG2 flow is compromised if the residual fraction is NG2-negative, which is the same failure as CD19 MRD after CD19-directed therapy.
Facts and figures
statementNG2/CSPG4 is absent from the candidate leukaemia-initiating fraction in t(4;11) ALL, so NG2-directed therapy debulks without eradicating and must be paired with an agent covering the CD34+CD19-CD117+ compartment.
originliterature-gap
prior probability0.45
predictions
[
  "In a KMT2A::AFF1 PDX treated with an NG2-directed agent, residual disease at nadir will be enriched for CD34+CD19-CD117+ cells relative to the pre-treatment sample.",
  "Relapse after NG2-directed therapy will be NG2-low or NG2-negative at first, and may regain NG2 as the bulk population re-expands, because NG2 is a malleable state rather than a clonal mark.",
  "Combining an NG2-directed agent with a menin inhibitor will produce deeper MRD responses than either alone, because the two act on different fractions."
]
experiments
[
  "exp-ng2-directed-agent-in-a-kmt2a-aff1-pdx-with-stem-fraction-readout"
]
analyses
[
  "ana-ng2-in-kmt2a-rearranged-patients-is-bimodal-and-the-552-fold-median-ratio-is-an-artefact"
]
falsifiers
[
  "SCOREABLE FALSIFIER: in a KMT2A::AFF1 patient-derived xenograft, sort CD34+CD19-CD117+ cells at diagnosis and measure surface NG2 by quantitative flow against an isotype control. If >= 30% of that fraction is NG2-positive above isotype, the reservoir argument fails. The 2006 series reports this fraction as uniformly negative in three t(4;11) patients, so a positive result in a modern series would overturn it directly.",
  "If NG2-directed treatment of a KMT2A::AFF1 PDX produces durable remission with no NG2-negative residual population detectable by flow at nadir, the hypothesis is refuted.",
  "If the CD34+CD19-CD117+ phenotype is shown not to contain the leukaemia-initiating cells in KMT2A::AFF1 ALL, the hypothesis is about the wrong compartment and must be rewritten rather than scored."
]
How this could bite us back
  • The entire hypothesis rests on three t(4;11) patients measured in 2006 with a stem-cell phenotype definition that has since been questioned. It is recorded as moderate confidence for that reason and should not be used to argue against an NG2 trial.
  • If NG2 is malleable, a paired therapy targeting the NG2-negative fraction faces the same problem in reverse, and the disease may simply oscillate between states faster than either agent acts.
  • High NG2 associates with worse event-free survival, so the NG2-high bulk is not harmless and debulking it may have real value even if it does not cure - reading this hypothesis as 'NG2 therapy is pointless' would be wrong.
Still open
  1. Has anyone repeated the 2006 measurement with modern markers for the leukaemia-initiating fraction in t(4;11) ALL? Three patients in 2006 is the entire evidence base for the central claim.
  2. If NG2 is a malleable state rather than a clonal property, what controls the switch? A drug that forced the NG2-negative fraction into NG2-positivity would convert this liability into a combination strategy.
  3. Does the NG2-negative fraction in the bimodal TARGET patients correspond to the CD34+CD19-CD117+ phenotype, or is it a different axis entirely?
Every source behind this entry (3)
  1. E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In six AML/abn(11q23) and three ALL/t(4;11) samples, candidate leukaemic stem-cell populations defined as CD34+CD38- (AML) and CD34+CD19-CD117+ (ALL) lacked NG2 expression in all patients, while NG2 correlated with the myeloid maturation marker CD33 on maturing cells.

    PMID 16643437DOI
    What this evidence cannot show

    Nine patients total, three of them t(4;11). Four-colour flow cytometry, 2006. The definition of a candidate stem-cell population by surface phenotype has been substantially revised since, and CD34+CD38- is no longer regarded as a reliable leukaemia-initiating fraction.

    Evidence that argues against this (1)
    • Serial xenotransplantation of primary MLL-AF4 leukaemias found NG2 to be a malleable marker not enriching for leukaemia-initiating cells - which means NG2-negativity does not define a stable reservoir either, and the clean separation this hypothesis assumes may not exist. Leukemia 2018
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Serial xenotransplantation of primary MLL-AF4-positive leukaemias indicated that NG2 is a malleable marker that does not enrich for leukaemia-initiating cells or CNS-infiltrating cells in MLL-rearranged infant B-ALL, while high NG2 expression associated with lower event-free survival, more circulating blasts and more frequent CNS disease in the Interfant cohort.

    PMID 28943635DOI
    What this evidence cannot show

    Abstract read in full. This paper carries a published correction, to an author's name spelling only, with no effect on the scientific content.

    Evidence that argues against this (1)
    • A marker that is malleable in serial xenografts may still be stably expressed under therapeutic pressure in a patient; xenograft passage is a strong selective and environmental perturbation that patients do not undergo. Leukemia 2018
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    In TARGET Pediatric ALL Phase II, CSPG4 is bimodal across MLL-positive samples: RNA-seq values of 0.01, 3.69, 39.65 and 49.17 RPKM against an MLL-negative median of 0.039 and maximum of 13.52.

    What this evidence cannot show

    n = 4. Bulk tumour expression cannot distinguish a uniformly intermediate population from a mixture of NG2-high and NG2-negative cells, which is precisely the distinction this hypothesis needs.

    Evidence that argues against this (1)
    • A patient at 0.01 RPKM may simply have low blast content in that sample rather than NG2-negative leukaemia, and no blast-percentage adjustment was applied. cBioPortal for Cancer Genomics 2026
Connected to