Audit of the infant presenting-feature evidence: what exists, what does not, what was rejected, what could not be read
also called infant presenting feature audit · bounded absence register infant ALL · Q7 enumeration
This is the working record behind the rebuilt infant clinical picture: every search that was run with its record count, every paper that looked relevant and was not used with the reason, and every paper that exists and could not be read. It is written so that someone repeating this work next year repeats the decisions, not just the searches.
The same thing, for a scientist
A register in three parts, run on 2026-09-19 against PubMed via NCBI E-utilities and Europe PMC. **Bounded absences**: queries that returned nothing, or returned only records that failed a title check, with the exact query strings. **Bounded rejections**: records that a reader would expect to be cited and that were read and not used, with the reason and what would change the judgement. **Bounded unreadables**: records that exist, are on topic, and could not be obtained by any route tried. The audit also records one **false absence** produced by this beat's own query construction, as a worked example, and two **recovered unreadables** where a previously-recorded unreadable proved fully available through a route the project had not used.
1. Bounded absences1 deeper
| # | Query (PubMed, 2026-09-19) | Records | After title check | Verdict |
|---|---|---|---|---|
| A1 | (haemoglobin[tiab] OR hemoglobin[tiab] OR platelet[tiab]) AND diagnosis[tiab] AND (infant[ti] OR infants[ti]) AND (acute lymphoblastic leukemia[ti] OR acute lymphoblastic leukaemia[ti]) | 0 | 0 | no published haemoglobin or platelet distribution at diagnosis in infant ALL |
| A2 | hepatomegaly[tiab] AND (infant[tiab] OR infants[tiab]) AND (leukemia[tiab] OR leukaemia[tiab]) AND KMT2A[tiab] | 0 | 0 | but see F1 - adding MLL recovers 10 records, so this is a symbol-set artefact, not an absence |
| A3 | (pallor[tiab] OR fatigue[tiab]) AND (infant[tiab] OR infants[tiab]) AND (leukemia[tiab] OR leukaemia[tiab]) AND (frequency[tiab] OR proportion[tiab] OR percent[tiab]) | 2 | 0 | a twin case report and an adult-survivor fatigue study |
| A4 | fever[tiab] AND (infant[ti] OR infants[ti]) AND (leukemia[ti] OR leukaemia[ti]) AND presentation[tiab] | 8 | 0 | six are infection complicating treatment, two are single-organ presentations |
| A5 | (infant[ti] OR infants[ti] OR infantile[ti]) AND (leukemia[ti] OR leukaemia[ti]) AND (presenting sign*[tiab] OR presenting symptom*[tiab] OR clinical presentation[tiab]) | 17 | 0 | 14 case reports, 3 reviews; no cohort reporting presenting-sign frequencies in infants |
| A6 | clinical presentation signs symptoms infant leukaemia first year of life | 18 | 0 | same picture |
The headline absence: there is no systematic review or pooled analysis of presenting signs and symptoms restricted to infants, analogous to Clarke et al. for ages 0-18. Every infant frequency this beat reports comes from a single cohort's baseline table.
F1: a false absence this beat produced and caught
Query A2 with KMT2A[tiab] returns 0. The same query with (MLL[tiab] OR KMT2A[tiab]) returns 10, including the TCCSG 62-infant cohort, the 144-case Chinese series and the neonatal leukaemia review - three of the most-used sources in this beat. Most of the infant organomegaly literature predates the KMT2A symbol.
Second example, from the same session: blinatumomab infant lymphoblastic leukemia Interfant pilot van der Sluis returns 0 records, and the paper it was looking for is PMID 37099340, which exists and is exactly on point. A natural-language string with an author name and a trial nickname is not a query. Any zero in this table would look identical to a real absence if the alias set or the syntax were wrong, and two of this beat's own zeros were.
2. Bounded rejections
| PMID | What it is | Why not used | What would change it |
|---|---|---|---|
| 42663567 | 153 infants <6 months with retinoblastoma, neuroblastoma or leukaemia, single centre, Iran; leukaemia 5-year OS 14.7% (14 of 17 died) | only 17 leukaemia cases, mixed with two solid tumours, and no presenting-feature frequencies reported | a leukaemia-only breakdown with presenting signs |
| 37066443 | Three infant B-ALL cases, Pakistan; all girls, all hyperleukocytic | three cases cannot estimate a frequency; used only as corroboration that hyperleukocytosis and hepatosplenomegaly are common | nothing; it is a case series by design |
| 30459160 | 'How I treat infant leukemia', Brown, Pieters, Biondi, Blood 2019 | an authoritative teaching review used as a map, not a source of measured frequencies | not applicable; correctly used as a map |
| 40016188 | Osteoclast targeting in high-risk B-ALL, Kotecha group | read in full; BCR::ABL1-positive and relapsed-adolescent models only, no infant, no KMT2A-rearranged data | a conditioned-medium or zoledronic acid experiment in infant KMT2A-r PDX |
| 36380143, 39941894, 33937032, 33837299, 35677155, 39985426, 42038742, 34706236 | eight Kotecha-group preclinical therapeutics and origins papers | out of scope for a clinical-presentation beat; enumerated so the next cycle does not treat them as undiscovered | any of them reporting a clinical presenting feature or an infant toxicity |
| 42215936, 41960026 | neonatal MPAL case report; MLL-AF4 twin case report | patient material belongs in patients/, which this beat does not own; cited only for a technical claim with no narrative | not applicable |
All of these would plausibly be expected in an infant-ALL bibliography. Each is here with a reason so that a future cycle can tell 'considered and dismissed' from 'never looked for'.
3. Bounded unreadables, and two recoveries1 deeper
Still unreadable on 2026-09-19:
| PMID | What it asserts | Routes tried |
|---|---|---|
| 41711432 | menin inhibitors display differential in vitro sensitivity in KMT2A-rearranged infant ALL (title only) | efetch XML; Europe PMC core; Europe PMC fullTextXML (no PMCID); Crossref works; publisher (HTTP 403) |
| 26369984 | outcome of relapsed infant ALL on Interfant-99 - the primary source for this graph's post-relapse prognosis | efetch XML (no abstract); Europe PMC core (no abstractText, not OA) |
Recovered - previously recorded as unreadable, read in full this cycle:
| PMID | PMC identifier | How |
|---|---|---|
| 42246204 | PMC13461956 | Europe PMC fullTextXML |
| 40016188 | PMC11868389 | Europe PMC fullTextXML |
Both genuinely have no abstract in PubMed or in the Europe PMC metadata record - one is a Letter to the Editor, the other a Blood Cancer Journal correspondence - so an abstract-level check reports 'unreadable' correctly and misleadingly. The rule this produces: before recording a paper as unreadable, query Europe PMC for a PMC identifier and try fullTextXML. A third paper used heavily by this beat, PMID 37752126, has no PubMed abstract either and was fully readable the same way; had the abstract check been the only test, this beat's best infant cohort table would have been missed.
A fourth trap, observed in a title
Reading ArticleTitle from efetch with a naive text accessor truncated PMID 41711432's title to 'Menin inhibitors display differential ' and PMID 41505662's to 'Sustained Benefit of Blinatumomab in Infants With ', and returned an empty string for PMID 40207717, whose title begins with an italicised gene symbol. The markup-loss failure the standing brief documents for abstracts occurs in titles too, and in the title case it can produce an empty field rather than a plausible sentence. Concatenating all descendant text nodes recovers the full title in all three cases.
4. Corrections owed to entries this beat does not own
Five entries in data/graph/nodes/clinical/ take presenting-feature frequencies from PMID 27647842, a meta-analysis of 33 cohorts of children aged 0-18 (n = 3084), and apply them to infants: concept-pallor-and-fatigue-anaemia, concept-fever-and-infection-at-presentation, concept-hepatosplenomegaly-infant-leukemia, concept-bruising-petechiae-thrombocytopenia and concept-leukemia-cutis-infant. The meta-analysis is correctly cited in each; the problem is scope, not sourcing. This beat has written infant-specific replacements or documented the absence for each feature and reports them for the coordinator rather than editing files it does not own.
| script | none; searches run interactively via NCBI E-utilities esearch/efetch/esummary and Europe PMC REST |
|---|---|
| status | run |
| date | 2026-09-19 |
| result summary | No infant-restricted study of presenting sign frequencies exists. Infant-specific figures are available for organomegaly, white cell count, CNS status and skin infiltration from single cohorts; none exists for pallor, anaemia, thrombocytopenia or fever at presentation. Two previously-unreadable papers were recovered via Europe PMC full-text XML; two remain unreadable. |
| bounded absences | 6 |
| false absences caught | 2 |
| bounded rejections | 13 |
| bounded unreadables | 2 |
| recovered unreadables | 2 |
inputs
[ "PubMed", "Europe PMC", "Crossref" ]
entries needing scope correction
[ "concept-pallor-and-fatigue-anaemia", "concept-fever-and-infection-at-presentation", "concept-hepatosplenomegaly-infant-leukemia", "concept-bruising-petechiae-thrombocytopenia", "concept-leukemia-cutis-infant" ]
- Would an Interfant, COG or JPLSG supplementary table release settle every absence in section 1 at once? The values are in their case-report forms.
- How many other 'unreadable' records across this graph are recoverable through Europe PMC full-text XML, given that two of five were?
- Should this graph adopt a standing rule that no paper is marked unreadable until a PMC identifier lookup and a fullTextXML attempt have both failed?
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database
PubMed searches restricted to infants return no cohort reporting the frequency of presenting signs and symptoms, and no haemoglobin or platelet distribution at diagnosis; the queries and counts are recorded above.
NCBI E-utilities esearch against PubMed: recorded queries and record counts, run 2026-09-19 · Q7-clinical-infant beat (this graph) · NCBI E-utilities 2026What this evidence cannot show
A counted absence is only as good as its symbol set and syntax, and two of this beat's own zeros were artefacts of exactly that (see F1). Non-English-indexed cohorts and values reported only in paywalled tables are invisible to these queries. Counts are dated 2026-09-19.
Evidence that argues against this (1)
- Query A2 returned 0 with KMT2A alone and 10 when MLL was added, which demonstrates within this very audit that a well-written zero can be wrong; every other zero here carries the same risk. Isoyama K et al. 1999
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
Two papers previously recorded in this graph as having no retrievable abstract are available in full through Europe PMC full-text XML (PMC13461956 and PMC11868389) and were read completely on 2026-09-19; a third heavily-used source, PMID 37752126, likewise has no PubMed abstract and was read in full at PMC10522653.
Europe PMC REST service: article and full-text-XML availability check, run 2026-09-19 · Q7-clinical-infant beat (this graph) · Europe PMC 2026What this evidence cannot show
This does not mean every unreadable record is recoverable; PMIDs 41711432 and 26369984 were checked the same way the same day and are not in PMC at all. The recovery route works for open-access correspondence in journals that deposit full text without depositing an abstract.
Evidence that argues against this (1)
- Recovering two of five attempts means the route helps but does not resolve the class, and the single most valuable of the five - the menin-inhibitor comparison - remains unavailable. Dymock SCC et al. 2026
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.review
The pooled source underlying five clinical entries in this graph is a meta-analysis of 33 cohort studies of children aged 0-18 (n = 3084) that reports hepatomegaly 64%, splenomegaly 61%, pallor 54%, fever 53%, bruising 52%, recurrent infections 49%, fatigue 46% and rash 35%, with no infant stratum.
Clinical presentation of childhood leukaemia: a systematic review and meta-analysis. · Clarke RT, Van den Bruel A, Bankhead C, Mitchell CD, Phillips B, Thompson MJ · Arch Dis Child 2016What this evidence cannot show
The meta-analysis is sound for its stated population. Its inclusion criteria were children aged 0-18 without date or language restriction, and it presents no age-stratified analysis, so no infant subgroup can be extracted from it.
Evidence that argues against this (1)
- Infant-only cohorts give materially different figures where they report the same sign - hepatosplenomegaly 70-88% against 64%/61%, CNS involvement 13.8-33.6% against 2-5% in older children - so the pooled figures are not conservative approximations for infants but different numbers. Mironova D et al. 2023
- E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database
A PubMed author query for Kotecha RS 2020-2026 returns 101 records, of which 24 carry infant, KMT2A or MLL in title or abstract, and eight preclinical therapeutics papers from that set were enumerated and deferred to other beats.
NCBI E-utilities esearch against PubMed: recorded queries and record counts, run 2026-09-19 · Q7-clinical-infant beat (this graph) · NCBI E-utilities 2026What this evidence cannot show
Author-name queries over-collect and were not disambiguated by ORCID or affiliation. The deferral judgement is this beat's, made from titles, and any of the eight could contain an infant toxicity or presenting-feature observation not visible in its title.
Evidence that argues against this (1)
- This beat's dispatch brief gives 69 papers for the same window against the 101 measured here, so the count itself is unstable and the deferral list should be re-derived rather than trusted. Karsa M et al. 2026
evidence audit for the infant clinical presentation beat
query A1 returns 0 records
queries A3-A6 return no infant cohort reporting sign frequencies
two of the three recorded unreadables are readable via Europe PMC fullTextXML
101 records 2020-2026, 24 infant/KMT2A, eight enumerated and deferred
- ← testsHypothesis: the spread in infant presenting-feature frequencies is mostly measurement, not biology
the audit supplies the variance the hypothesis tries to explain