Children's Oncology Group — the trial network, the AALL protocol series and the biobank that no single lab can substitute

Self-description by the organisation on its own website; institution and investigator counts are unaudited and the 'largest repository' characterisation is a claim by an interested party. The NCTN and PEP-CTN memberships are externally checkable facts about NCI programme structure and are the load-bearing part of this citation. Retrieved 2026-09-18.

• Qualifies 'no single lab can substitute' by naming what COG is not. COG is the North American network; it is not the route to newly diagnosed infants, and it is not the only paediatric menin-inhibitor route. Two registry records establish this. Interfant-21 (NCT05327894), sponsored by the Princess Maxima Center with the University of Milano-Bicocca and Amgen Europe as collaborators, is RECRUITING with a planned enrolment of 160 infants under one year with KMT2A-rearranged ALL or MPAL, started December 2022, primary completion September 2027 -- the frontline infant population, outside COG entirely. TINI 2 (NCT05848687), sponsored by Tanja A. Gruber with POETIC, Amgen and Kura Oncology as collaborators, is RECRUITING with a planned enrolment of 90 newly diagnosed infants at 24 North American sites and includes ziftomenib dose-finding -- a menin-inhibitor infant study that is not a COG protocol and runs at many non-COG-exclusive centres. So a lab that treats COG as the single gateway to infant material will miss the only frontline infant cohorts and the only frontline infant menin-inhibitor experience. Caveat: registry records establish that these studies exist, are recruiting and are sponsored elsewhere; they do not establish that their biospecimens are more accessible than COG's, and both are single-sponsor academic protocols with their own governance. Read: registry records only. Princess Maxima Center for Pediatric Oncology; University of Milano Bicocca; Amgen Europe B.V. 2026

This paper is about osteosarcoma, so the 15,000-specimen figure is disease-specific and says nothing about the size of the leukaemia holdings. It establishes the existence, location and governance philosophy of the central bank, not the inventory relevant to infant ALL. It is also from 2017; holdings and infrastructure will have changed. It does not describe the specimen application process. According to PubMed; DOI https://doi.org/10.1371/journal.pone.0181204

• Scale is not access. An inventory of that size describes what the organisation holds; what an external investigator can obtain depends on membership, on a study-specific request, on a material transfer agreement and on whether the particular samples — infant, KMT2A-rearranged, with viable cells or extractable nucleic acid — still exist in usable quantity. Those are different questions and the second is the one a plan depends on. The osteosarcoma inventory figure is also the wrong disease: it demonstrates that a large bank exists, not that infant KMT2A-rearranged material is in it at a useful depth. Cancer Discovery 2021

Registry records state protocol design and enrolment targets, not results; AALL2121 has reported nothing and 'active, not recruiting' means accrual closed, not that data exist publicly. The 1-month-to-6-years range is wider than infancy strictly defined, so the cohort mixes true infants with young children. Whether a specific specimen set is available for outside analysis is not addressed by the registry. Records retrieved 2026-09-18.

• This is the entry's central strategic claim and the protocol record undercuts it in three specific ways. The entry's partnering_angle says 'A laboratory arriving with a validated assay for menin-inhibitor response or escape in KMT2A-rearranged infant blasts is describing the correlative analysis that cohort exists to support', and its third open question asks whether the correlative analyses have already been allocated. The registered protocol answers that: they are already written in. AALL2121's exploratory outcome measures include 'Changes in KMT2A-R transcriptional program' (RNA from cryopreserved leukaemic cells, RNA-seq, principal component analysis and gene-set enrichment on paired pre- and post-treatment samples, correlated with clinical response, pharmacokinetics and menin displacement) and 'Changes to menin displacement' (paired pre/post samples tested for menin occupancy by CUT&RUN, with histone modifications and protein complexes by CUT&Tag, correlated with clinical response). Those are precisely the assays a mechanism lab would propose, and they are protocol-specified with a five-year reporting window. Second, the primary MRD endpoint is restricted to 'patients age >= 6 months at enrollment'; patients under 6 months appear only as a descriptive exploratory outcome, and eligibility runs to under 6 years with diagnosis under 2 years -- so the true under-6-month infant subgroup is not powered and may be very small, which bears directly on this entry's fourth open question. Third, the protocol excludes patients previously treated with revumenib, so this cohort cannot be a source of post-revumenib resistance material. Status as at 2026-09-18: ACTIVE_NOT_RECRUITING, enrolment 78, primary completion 2027-12-31. The honest revision is that the remaining opening is an orthogonal assay the protocol does not list -- non-menin dependency, breakpoint-resolved analysis, condensate or localisation readouts -- not the menin-displacement work, which is taken. Caveat: registry outcome measures describe planned analyses, not who will perform them or whether specimen aliquots remain; a COG investigator could still confirm capacity. Read: full registry record including detailed description, eligibility and outcome measures, retrieved 2026-09-18. Children's Oncology Group 2026

Randomisation was terminated early on data and safety monitoring committee recommendation with 80 of 131 planned events, so the trial was underpowered for its primary endpoint and did not reach statistical significance for disease-free survival. Cited here as evidence of COG's operational scale and its ability to run randomised comparisons in relapsed paediatric ALL, not as evidence about blinatumomab. According to PubMed; DOI https://doi.org/10.1001/jama.2021.0669

• AALL1331 enrolled patients aged 1 to 30 with first relapse of B-ALL, which excludes infants under one year — the population this graph is about. Continental scale in paediatric leukaemia does not translate into scale in infant leukaemia, where the whole disease is a few hundred cases a year worldwide. Leukemia 2021

A retrospective model built on accumulated trial data, so the population is trial-enrolled children rather than all children with B-ALL, and the model's performance in subgroups as rare as KMT2A-rearranged infant ALL is not established by the headline figure. The 21,199 number describes the modelling cohort, not the size of any single accessible dataset. According to PubMed; DOI https://doi.org/10.1038/s41375-024-02166-1

• A prognostic index developed and validated on 21,199 B-ALL patients is a model for the population it was fitted to, and infant ALL is treated on separate protocols and is largely absent from frontline COG B-ALL trials. A model that has barely seen the beat's patients cannot be assumed to calibrate on them. Statistics in Medicine 2006

Explicitly excluded infant ALL, BCR-ABL1-positive and hypodiploid ALL from the discovery cohort, so it says nothing directly about infant KMT2A-rearranged disease. Cited as evidence that COG-derived cohorts enable discovery an individual laboratory could not make, not for its biological conclusion. Array-based copy-number technology of 2009. According to PubMed; DOI https://doi.org/10.1056/NEJMoa0808253

• The IKZF1 discovery was made in high-risk B-progenitor ALL, which is not KMT2A-rearranged infant disease, and IKZF1 deletion is uncommon in KMT2A-rearranged cases. It demonstrates what the cohorts can support in general and is not evidence that this specific question can be answered in them. Leukemia 2021