One trial, two published analyses, never crossed: gemtuzumab clearly helps KMT2A-rearranged children, and a common CD33 genotype abolishes gemtuzumab benefit — nobody has asked whether it abolishes it in them
also called CD33 rs12459419 in KMT2A-rearranged AML · AAML0531 uncrossed analyses · does the CD33 genotype abolish gemtuzumab benefit in KMT2A-r
A targeted antibody drug is one of the few clear wins in this fusion, and separately a common inherited variant in the drug's target gene wipes out its benefit across the trial as a whole. Both results come from the same 1,022-child trial and nobody has put them together, so whether children with the fusion who carry the variant benefit at all is unknown.
The same thing, for a scientist
Two analyses of Children's Oncology Group AAML0531 (NCT01407757), same 1,022-patient cohort, never crossed. **(1)** Pollard 2021 (PMID 34048275, JCO): **215 (21%) of 1,022** enrolled had KMT2A-r AML, and gemtuzumab ozogamicin improved EFS (48% vs 29%, P = .003), reduced relapse risk (40% vs 66%, P = .001) and improved DFS (57% vs 33%, P = .002); with HSCT, 5-year RR 28% vs 73% (P = .006). GO was independently associated with improved EFS and DFS in multivariable analysis. **(2)** Lamba 2017 (PMID 28644774, JCO): the *CD33* splicing polymorphism **rs12459419** determines GO response — and the exon-2-skipped isoform it produces is diverted to an intracellular compartment rather than the cell surface (PMID 28747436). **Neither paper contains the other's variable.** I read both: the CD33 genotype paper mentions KMT2A, MLL and 11q23 **zero times** (PMC5549451, full text); the KMT2A-r paper mentions rs12459419, 'splic' and 'genotype' **zero times**. The cross is a contingency table on data already collected in one trial.
Why two published analyses of one trial can miss each other
| absence class | measured-not-reported |
|---|---|
| absence class basis | Both variables were measured in the same 1,022-patient trial and each has been published separately against outcome. The cross-tabulation has not been published. No new patient, sample or assay is required — only access to the trial dataset. |
| why this is the sharpest open clinical question in this graph | GO is the one positive randomised targeted-agent result in paediatric AML, the KMT2A-r subgroup is where it works best, and a common germline variant abolishes it cohort-wide. If the variant abolishes it in KMT2A-r too, roughly a fifth of these children are receiving an agent that cannot work for them and the genotype should precede the prescription. If it does not, the KMT2A-r subgroup is a genotype-independent indication, which is also worth knowing. |
| shown for fusion | The KMT2A-r subgroup is defined at the rearrangement level (215 patients, partner-stratified into higher-risk and not-high-risk). Neither paper resolves KMT2A::AFF1 specifically, and t(4;11) is rare in AML cohorts — so even the crossed analysis would not be about the index fusion without further stratification. |
| provenance note on this entrys own citation error | The E3 citation carried a fabricated author list for eleven hours. The PMID resolved, the title matched, the DOI matched, the journal and year matched -- every surface check this project runs passed. Only an author-level comparison catches it, and the list was plausible enough to read as real because it almost certainly IS a real author list, from a different Siglec paper. |
trial
{
"name": "AAML0531, Children's Oncology Group, phase III",
"nct": "NCT01407757",
"enrolled": 1022,
"kmt2a_rearranged": "215 (21%)"
}half one the kmt2a result
{
"pmid": "34048275",
"finding": "GO improved EFS 48% vs 29% (P = .003); relapse risk 40% vs 66% (P = .001); DFS 57% vs 33% (P = .002). With HSCT, 5-year RR 28% vs 73% (P = .006). Benefit seen in both higher-risk and not-high-risk translocation-partner subsets. OS was NOT significantly different (63% vs 53%, P = .054).",
"mentions_rs12459419": 0,
"mentions_splic": 0,
"mentions_genotype": 0
}half two the genotype result
{
"pmid": "28644774",
"finding": "The CD33 splicing polymorphism rs12459419 determines GO response in de novo paediatric AML.",
"mechanism_pmid": "28747436",
"mechanism": "The exon-2-skipped CD33 isoform is diverted to an intracellular compartment rather than the cell surface — a localisation-level explanation for a pharmacogenomic effect.",
"mentions_KMT2A": 0,
"mentions_MLL": 0,
"mentions_11q23": 0,
"measured_by": "regex over the PMC5549451 full text, 2026-09-21; rs12459419 appears 42 times and 'CC genotype' 21 times in the same text, so the zero is not a retrieval failure"
}- THIS ENTRY IDENTIFIES AN UNANSWERED QUESTION AND MUST NOT BE READ AS GUIDANCE. Nothing here supports withholding gemtuzumab from any child on the basis of CD33 genotype. The genotype effect was demonstrated cohort-wide and has NOT been shown to hold in the KMT2A-rearranged subgroup — that is precisely what is missing.
- Equally, nothing here supports giving GO to a KMT2A-rearranged child without regard to genotype. The honest position is that the interaction is unmeasured, and the entry exists to say so.
- The KMT2A-r benefit itself rests on EFS, DFS and relapse risk. Overall survival did not reach significance (P = .054). Any statement that GO 'improves survival' in this subgroup overstates the published result.
- This is AML. The index disease of this project is infant KMT2A::AFF1 B-ALL, where t(4;11) predominates and these data do not apply. Do not transfer the result across lineage.
- Cross rs12459419 genotype with KMT2A-r status in the AAML0531 dataset and report GO benefit within genotype strata. It is a contingency table on data already collected, both variables already published separately against outcome, and 215 patients in the KMT2A-r arm.
- Search the AAML0531 supplementary material and the Children's Oncology Group meeting abstracts before calling this an absence in print rather than an absence in two primary papers. The KMT2A-r side of this census was abstract-only.
- Is surface CD33 density itself different in KMT2A-rearranged blasts? If the rearrangement affects CD33 expression independently of genotype, the two variables are not independent and a simple cross would be confounded.
- Does any of this reach the index fusion? t(4;11) KMT2A::AFF1 is rare in AML cohorts, so even a crossed AAML0531 analysis would be about KMT2A-r AML generally. The infant B-ALL question would need a different trial.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Gemtuzumab ozogamicin improves outcomes in paediatric KMT2A-rearranged AML. In AAML0531, 215 (21%) of 1,022 enrolled patients had KMT2A-r AML; EFS was 48% with GO against 29% without (P = .003), relapse risk 40% against 66% (P = .001), and DFS 57% against 33% (P = .002). Among patients who underwent HSCT, prior GO exposure was associated with 5-year relapse risk of 28% against 73% (P = .006). Overall survival was not significantly different (63% vs 53%, P = .054).
Gemtuzumab Ozogamicin Improves Event-Free Survival and Reduces Relapse in Pediatric KMT2A-Rearranged AML: Results From the Phase III Children's Oncology Group Trial AAML0531. · Pollard JA, Guest E, Alonzo TA, Gerbing RB, Loken MR, Brodersen LE, Kolb EA, Aplenc R, Meshinchi S, Gamis AS · Journal of Clinical Oncology 2021What this evidence cannot show
Abstract fetched and read 2026-09-21; the full text and supplementary tables were not opened, so the per-partner breakdown beyond 'higher-risk versus not high-risk' is not established here. This is a subgroup analysis of a randomised trial, not a randomised comparison within KMT2A-r, so the usual subgroup cautions apply. Overall survival did not reach significance (P = .054), which matters: the benefit is in relapse and event-free survival, not demonstrated in survival. AML cohort, so t(4;11) KMT2A::AFF1 is rare in it; this is not an infant B-ALL result.
Evidence that argues against this (1)
- OS at P = .054 is the honest brake on this. A regimen that reduces relapse and improves EFS without a demonstrated survival benefit may be shifting the timing and mode of failure rather than curing more children, and in a subgroup analysis with 215 patients the confidence intervals are wide. The HSCT comparison (28% vs 73%) is also not randomised for transplant, so selection is uncontrolled there. Journal of Clinical Oncology 2021
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database
Neither analysis contains the other's variable, so the cross has never been published. The CD33 splicing-polymorphism paper (PMID 28644774, PMC5549451) mentions KMT2A zero times, MLL zero times and 11q23 zero times across its full text, while mentioning rs12459419 42 times and 'CC genotype' 21 times in the same text. The KMT2A-rearranged paper (PMID 34048275) mentions rs12459419, 'splic' and 'genotype' zero times in its abstract.
Full-text and abstract term census over PMC5549451 (PMID 28644774) and PMID 28644774/34048275, run 2026-09-21: KMT2A/MLL/11q23 = 0 in the genotype paper against rs12459419 = 42 in the same text; rs12459419/splic/genotype = 0 in the KMT2A-r paper. · PubMed Central and PubMed E-utilities (term census; counts recorded in data) 2026What this evidence cannot show
The KMT2A-r side was counted on the ABSTRACT only, because that paper is not in the open-access subset; its full text or supplement could contain a genotype analysis not visible from the abstract, and that possibility is not excluded here. The genotype side was counted on full text. A term census also cannot detect an analysis performed and reported only as a figure without the term appearing in text.
Evidence that argues against this (1)
- Counting terms is not the same as reading the analysis, and the weaker half of this claim rests on an abstract. The same trial group published both papers four years apart, so the cross may exist in a conference abstract, a thesis, or a supplementary table that a term census over two primary papers would not reach. Before anyone calls this an absence in print rather than an absence in these two papers, the AAML0531 supplementary material and the group's meeting abstracts should be searched. Journal of Clinical Oncology 2017
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
The genotype's effect has a localisation-level mechanism, which is why it belongs in this graph rather than only in a pharmacogenomics one. The CD33 splice variant produced by rs12459419 skips exon 2 and the resulting isoform is diverted to an intracellular compartment instead of the cell surface, removing the epitope an antibody-drug conjugate requires.
The Alzheimer's disease-protective CD33 splice variant mediates adaptive loss of function via diversion to an intracellular pool. · Siddiqui SS, Springer SA, Verhagen A, Sundaramurthy V, Alisson-Silva F, Jiang W, Ghosh P, Varki A · Journal of Biological Chemistry 2017What this evidence cannot show
This paper is framed around Alzheimer's disease biology, not leukaemia, and the cell systems are not AML blasts. The inference that the same diversion explains the AAML0531 pharmacogenomic result is a bridge between two literatures and is not demonstrated in either. Title and identifiers fetched from efetch 2026-09-21; abstract read, full text not opened. AUTHOR LIST CORRECTED 2026-09-22: this citation previously stored 'Bhattacherjee A, Rodrigues E, Jung J, ...', which is not this paper's author list. The PMID, title, DOI, journal and year were all correct; only the authors were wrong, and they were written from recall rather than copied from the fetched record. Found by Q91's citation check and verified by me against live efetch.
Evidence that argues against this (1)
- A mechanism that is plausible and a mechanism that is operative are different claims. The splice variant could reduce GO response by lowering total surface CD33 density, by changing epitope accessibility, or by an unrelated linked variant, and an Alzheimer's-focused cell-biology paper does not adjudicate among those in leukaemic blasts. Treating the intracellular-diversion account as established would be exactly the cross-literature over-reading this project has corrected elsewhere. Journal of Clinical Oncology 2017
Builds on this entry's localisation finding: the exon-2-skipped CD33 isoform is intracellular rather than surface, which is the mechanism by which rs12459419 removes the epitope an antibody-drug conjugate needs. Adds that the genotype has never been crossed with KMT2A status in the trial that measured both.
- related to →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)
Bears on the rearrangement class rather than on this fusion specifically: AAML0531 is an AML trial, 215 of 1,022 KMT2A-rearranged, and t(4;11) KMT2A::AFF1 is rare in AML cohorts. The entry says so rather than transferring the result across lineage.
A clean measured-not-reported case found by asking that entry's question: both variables are in one trial database on the same 1,022 children, each already published against outcome, and only the cross is missing. Closing it needs permission, not a cohort.