The protocol and era confound: the partner datasets span 1983 to 2016 and outcome moved under them
also called era confound KMT2A partner outcome · protocol heterogeneity in partner cohorts
The studies comparing partner genes were assembled from patients treated over three decades on dozens of different regimens. Treatment improved a great deal in that time, so some of what looks like a partner difference is a difference in what year and what country a child was treated.
The same thing, for a scientist
Every partner-outcome dataset in this beat is retrospective and pooled across protocols. The 497-patient childhood 11q23 ALL analysis covers 1983-1995 across 11 groups; the 756-patient 11q23 AML series pools 11 collaborative groups; the 629-patient Ponte-di-Legno ALL cohort covers 1995-2010 across 17 groups; the 1,130-patient I-BFM AML cohort covers 2005-2016 across many national protocols. Outcome moved substantially over that span and it moved unevenly by region: Interfant-06 itself reports that West European and North American groups achieved 6-year EFS and OS 10-12 percentage points higher than other participating countries within the same protocol. A partner enriched in one country, one group or one decade therefore inherits that group's outcome. None of the published partner analyses reports an era-adjusted or group-adjusted partner effect.
The within-protocol regional gap is the measurable version of this
Interfant-06 ran one protocol and still saw a 10-12 point EFS and OS gap between participating regions 1. That is a lower bound on how much non-biological variation a pooled multi-group partner comparison carries, and it is larger than several of the partner effects this beat located. It is also the cleanest available estimate, because protocol is held fixed.
Era is not just noise, it changes what can be measured
Partner assignment moved from karyotype to FISH to targeted RT-PCR to RNA sequencing over the same span. Each method has a different blind spot, so the set of patients assigned to 'other' or 'unknown' changes with era, and the denominator of every named partner changes with it. An era effect in this literature is simultaneously a treatment effect and an ascertainment effect, and no published analysis separates them.
What would separate it
Fit a partner effect within a single protocol with adequate n. Two candidates exist: COG AAML1831 (NCT04293562), which has adopted partner-based initial risk stratification and therefore records partner prospectively; and Interfant-21 (NCT05327894), if it pre-specifies partner reporting. A within-protocol partner estimate has never been published in this disease.
| confound | protocol and treatment era |
|---|---|
| era adjusted partner effect published | false |
| within protocol partner estimate published | false |
| secondary effect | partner assignment method changed with era (karyotype to FISH to RT-PCR to RNA-seq), so era is simultaneously a treatment and an ascertainment confound |
| source | PMIDs 31283407, 36256911, 36996387, 12057554 |
within protocol regional gap
{
"trial": "Interfant-06",
"pmid": "31283407",
"west_eu_na_efs_6y_pct": 49.4,
"west_eu_na_os_6y_pct": 62.1,
"whole_trial_efs_6y_pct": 46.1,
"whole_trial_os_6y_pct": 58.2,
"stated_gap_pct": "10 to 12",
"title": "Outcome of Infants Younger Than 1 Year With Acute Lymphoblastic Leukemia Treated With the Interfant-06 Protocol: Results From an International Phase III Randomized Study.",
"authors": "Pieters R, De Lorenzo P, Ancliffe P, et al.",
"venue": "J Clin Oncol",
"year": 2019,
"title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
}dataset spans — the numbers behind the chart
[
{
"pmid": "12057554",
"n": 497,
"years": "1983-1995",
"groups": "11 study groups and single institutions",
"title": "Outcome of treatment in childhood acute lymphoblastic leukaemia with rearrangements of the 11q23 chromosomal region.",
"authors": "Pui CH, Gaynon PS, Boyett JM, et al.",
"venue": "Lancet",
"year": 2002,
"title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
},
{
"pmid": "19528532",
"n": 756,
"groups": "11 collaborative groups",
"title": "Novel prognostic subgroups in childhood 11q23/MLL-rearranged acute myeloid leukemia: results of an international retrospective study.",
"authors": "Balgobind BV, Raimondi SC, Harbott J, et al.",
"venue": "Blood",
"year": 2009,
"title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
},
{
"pmid": "36256911",
"n": 629,
"years": "1995-2010",
"groups": "17 Ponte-di-Legno members",
"title": "Outcomes of Childhood Noninfant Acute Lymphoblastic Leukemia With 11q23/KMT2A Rearrangements in a Modern Therapy Era: A Retrospective International Study.",
"authors": "Attarbaschi A, Möricke A, Harrison CJ, et al.",
"venue": "J Clin Oncol",
"year": 2023,
"title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
},
{
"pmid": "36996387",
"n": 1130,
"years": "2005-2016",
"groups": "many national protocols",
"title": "Measurable Residual Disease and Fusion Partner Independently Predict Survival and Relapse Risk in Childhood KMT2A-Rearranged Acute Myeloid Leukemia: A Study by the International Berlin-Frankfurt-Münster Study Group.",
"authors": "van Weelderen RE, Klein K, Harrison CJ, et al.",
"venue": "J Clin Oncol",
"year": 2023,
"title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
}
]separating designs
[ "partner effect estimated within a single protocol (COG AAML1831, NCT04293562)", "partner reporting pre-specified in Interfant-21 (NCT05327894)", "era and study group as covariates in a pooled individual-patient model" ]
- A menin-inhibitor trial comparing its results with historical partner-stratified controls will inherit both the era and the regional confound, and the resulting apparent partner-specific benefit could be entirely geographic.
- Because partner ascertainment improved with era, modern cohorts contain more named rare partners, so the 'other' arm shrinks over time and its composition changes - a comparison of rare-partner outcome across eras is comparing different sets of patients.
- Does partner distribution differ by participating country within Interfant-06 and within the I-BFM AML cohort? If it does, the regional outcome gap contaminates the partner comparison directly.
- What is the partner effect estimated within AAML1831 alone, where protocol and era are fixed and partner is recorded prospectively?
- Can the ascertainment component of the era effect be bounded by re-genotyping banked samples from an older cohort with RNA-seq?
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Within a single protocol, Interfant-06, 6-year EFS and OS in West European and North American groups were 49.4% (SE 2.5) and 62.1% (SE 2.4), which the authors state were 10% to 12% higher than in other participating countries, against whole-trial figures of 46.1% and 58.2%.
Outcome of Infants Younger Than 1 Year With Acute Lymphoblastic Leukemia Treated With the Interfant-06 Protocol: Results From an International Phase III Randomized Study. · Pieters R, De Lorenzo P, et al. · J Clin Oncol 2019What this evidence cannot show
The regional comparison is descriptive and was not adjusted for case mix, so part of the gap could be patient characteristics rather than care delivery. The regional grouping is coarse. Whether partner distribution differs by region within the trial is not reported.
Evidence that argues against this (1)
- JPLSG MLL-10 in Japan reported 3-year EFS of 66.2% for KMT2A-rearranged infants, well above the Interfant-06 whole-trial figure, which cuts against a simple West-versus-rest reading of the regional gap. Tomizawa D et al. 2020
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The partner-outcome datasets are pooled across many groups and long spans: 497 patients treated 1983-1995 across 11 groups and institutions; 756 patients across 11 collaborative groups; 629 patients across 17 groups covering 1995-2010; 1,130 patients across many national protocols covering 2005-2016.
Outcomes of Childhood Noninfant Acute Lymphoblastic Leukemia With 11q23/KMT2A Rearrangements in a Modern Therapy Era: A Retrospective International Study. · Attarbaschi A, Möricke A, et al. · J Clin Oncol 2023What this evidence cannot show
The spans and group counts are taken from each study's own description; none of the four reports an era-adjusted or group-adjusted partner effect, so the magnitude of the confound in each is unquantified rather than quantified and dismissed.
Evidence that argues against this (1)
- The I-BFM study's partner risk group survived adjustment for end-of-induction MRD, a variable that captures treatment response and therefore indirectly captures some protocol intensity, which argues the partner effect is not purely an era artefact. van Weelderen RE et al. 2023
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Prognosis for non-infant childhood KMT2A-r ALL improved relative to historical data over the 1995-2010 span (5-year EFS 69.1% +/-1.9 in the modern cohort), explicitly stated by the authors as an improvement compared with historical data.
Outcomes of Childhood Noninfant Acute Lymphoblastic Leukemia With 11q23/KMT2A Rearrangements in a Modern Therapy Era: A Retrospective International Study. · Attarbaschi A, Möricke A, et al. · J Clin Oncol 2023What this evidence cannot show
Historical comparison rather than a within-cohort era analysis; improvement is attributed to therapy generally and not decomposed.
Evidence that argues against this (1)
- Interfant-06 reported that outcome for infant ALL did not improve compared with Interfant-99, so the era improvement is age-specific and cannot be assumed across all the pooled partner datasets. Pieters R et al. 2019
- related to →Interfant-06 — 651 infants, and the definitive answer that myeloid-style consolidation does not help
a 10-12 point EFS and OS gap between regions within one protocol
- related to →756 children with 11q23/MLL-rearranged AML: 5-year EFS runs from 11% to 92% depending on the partner
partner effect estimated by pooling 11 collaborative groups with no era adjustment
a single-centre single-protocol cohort, which removes the pooling confound
- ← depends onThe analysis that would settle it: a partner-by-lineage interaction term in pooled individual-patient data
era and study group must be covariates or the estimate is uninterpretable