Oncofusion

start / graph / ana-q198-all-three-withdrawn-menin-inhibitor-trials-admit-infants-a-post-hoc-pattern-whose-stated-reasons-are-business-and-administrative

Analysisweak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.

All three withdrawn mechanism-targeted trials admit infants and none of the 52 trials that do not has been withdrawn (exact p = 0.0032, noticed after looking), but the stated reasons are business and administrative and all three are 2025-2026 registrations

Three menin-drug trials were cancelled before enrolling anyone, and all three would have been open to babies. It is probably a coincidence of timing and company decisions, but it is worth watching.

The same thing, for a scientist

POST HOC. Of the 10 infant-admitting mechanism-targeted trials, 3 were withdrawn (n=0) and 0 terminated; of the 52 that do not admit infants, 0 were withdrawn and 4 terminated. P(all 3 withdrawn fall among the 10 infant-admitting of 62) = 0.0032. Reasons: 'Business Decision', 'PI requested', 'Protocol moved to Withdrawn'. Start dates 2025-12, 2026-05, 2026-09.

Go deeper
The three
NCTmin agestartverbatim reason
NCT0552108730 Days2025-12Business Decision
NCT070529946 Months2026-05PI requested
NCT0749846530 Days2026-09Other - Protocol moved to Withdrawn
Facts and figures
scriptanalyses/q198-the-registry-of-stopped-trials/analyze.py
p one sided exact0.0031729243786356425
post hoctrue
infant admitting t1
terminated0.000terminated: 0withdrawn3.000withdrawn: 3n10.0n: 10
non infant t1
withdrawn0.000withdrawn: 0terminated4.000terminated: 4n52.0n: 52
infant admitting t1 — the numbers behind the chart
{
  "n": 10,
  "withdrawn": 3,
  "terminated": 0
}
non infant t1 — the numbers behind the chart
{
  "n": 52,
  "withdrawn": 0,
  "terminated": 4
}
How this could bite us back
  • If paediatric and infant protocols are systematically the ones sponsors withdraw first, infants lose access to the drugs whose evidence base already excludes them. This entry cannot show that - it only flags it.
Still open
  1. Re-run on the registry in 12 months: does the split persist among trials that have had time to enrol, or is it an artefact of recency?
Every source behind this entry (1)
  1. E1weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.database

    All three withdrawn mechanism-targeted trials (NCT05521087, NCT07052994, NCT07498465; n=0 each) admit infants under one year, whereas none of the 52 mechanism-targeted trials that do not admit infants was withdrawn; the exact one-sided probability of that split by chance is 0.0032.

    What this evidence cannot show

    This pattern was NOTICED after reading the data, so the p-value is descriptive and not a test: with several ways to cut the same 62 trials, some split will look this lopsided. Exact hypergeometric calculation, one-sided.

    Evidence that argues against this (1)
    • Three confounds, any of which could produce this alone. (1) All three are 2025-2026 registrations, and protocols registered recently are withdrawn before enrolment far more often than old ones. (2) The stated reasons are business, PI request and administrative, not anything about infants. (3) n=3; one more withdrawal in the other group removes the pattern. ClinicalTrials.gov 2026
Connected to