April 2023 — Monthly Briefing¶
1. What was read¶
729 research articles from 33 of the 33 journals on the roster.
A further 212 records (23% of the 941 this month's harvest returned) carry no abstract and are not research: news, correspondence, editor highlights and obituaries. PubMed types them Journal Article and nothing else, so the publication-type filter cannot see them — across the whole index they are 22.7% of it and 35% of general, and 3,801 of them are JAMA. Every count below is over the full harvest and is inflated by this; the figure above is not.
| General | Cancer | Immune | Bioinfo | Sysbio |
|---|---|---|---|---|
| 431 | 169 | 66 | 246 | 29 |
Largest contributors: Nature 111, Science 87, JAMA 86, Nucleic Acids Res 75, Bioinformatics 59, Lancet 49.
2. Net change on the thread curve¶
| Thread | This month | Cumulative | Status |
|---|---|---|---|
cell-therapy |
+9 / −1 | +31 / −10 | forming |
ctdna-mrd |
+2 | +11 / −8 | splitting |
foundation-models |
+2 / −1 | +8 / −8 | splitting contested → splitting |
immunotherapy-resistance |
+11 / −3 | +40 / −13 | forming |
interception |
+13 | +27 / −3 | forming |
method-credibility |
−13 | +0 / −50 | crisis |
neoantigen-vaccines |
+2 / −2 | +12 / −5 | forming |
neuro-immune-microbiome-axis |
+10 / −1 | +26 / −4 | forming |
next-gen-checkpoints |
+2 | +9 / −5 | splitting |
single-cell-3d-genome |
+2 | +9 / −5 | splitting |
suppressive-population-inversion |
+4 / −1 | +4 / −13 | crisis |
Status is derived by rule, not asserted. Fewer than 5 cards → too-early; then refutes ≥ 2×supports → crisis; refutes > supports → contested; supports ≥ 2×refutes → forming; else splitting.
3. The evidence that moved it¶
A note that governs every claim below, and every claim in May and June. This quarter was first read against titles and journals only, and re-read in full from abstracts on 2026-09-04. The older prose below was written under the first reading and still says, in places, that a title cannot carry an effect size or a control — true, and no longer a limit on what is carded. 38% of April's cards state something the title does not. Where the argument still turns on something only a title supports, it says so rather than asserting it.
The microbiome arrives at three evidence grades in the same month, from three
independent groups. 37028428
(Cell) names an organism and the dietary tryptophan metabolite it releases as
required for checkpoint inhibitor efficacy.
37053311 (Science) engineers skin
commensals into an antitumour agent — the same biology built as a drug rather
than observed as a correlate. 36914893
(Nat Med) reads the effect in patients receiving CD19 CAR-T, with antibiotic
exposure separated out so the microbial signal survives the obvious confounder.
Mechanism, manufactured agent, and human correlate, in four weeks, in three
journals. That is the strongest structure literature alone can offer, and these
three are the core of the +10 neuro-immune-microbiome-axis takes this month.
A second convergence, on the antigen space, and it is not confined to this
month. 36973455 reads
transposable elements pan-cancer as a source of tumour-specific antigens;
37012458 reaches the same
conclusion from chimeric TE–gene transcripts — same journal, same month,
different groups. The corpus already holds
36735774 (Sci Immunol, February)
on noncanonical exon–TE splice junctions as immunogenic neoantigens. Three
groups across two months are enlarging the neoantigen-vaccines target space
away from point mutations, which is the part of the thread's claim — that the
space is large enough — that mutational burden alone has never settled.
The neural arm gets an intervention rather than an observation. 37099632 (Sci Transl Med): β-blockade changes what an anthracycline does to metastasis. The thread asserts that neurotransmitters are a shared currency between the neural, microbial and lymphoid arms; a currency you can spend to alter the outcome of an existing cytotoxic drug is a materially stronger form of that claim than one you can only measure. The title does not give the effect size, and this is a mouse result.
Engineered cell therapy leaves the B cell, at the highest evidence grade the
thread can reach. 37018492
(NEJM): a GD2-directed CAR-T in relapsed high-risk neuroblastoma — a solid
tumour, in children, against a non-CD19 antigen, in the journal that publishes
registrational readouts. cell-therapy is +9/−1 this month, and this is the top
of that supporting side rather than the whole of it.
Corrected 2026-09-04
This read "cell-therapy is +2/−1 this month and this is the supporting
half", which made the NEJM trial one of two supports and gave it half the
month's weight. The abstract re-scan puts the month at nine supports —
manufacturing and editing routes, a stem-like CAR-T persistence result, a
second-line lisocabtagene readout, real-world idecabtagene data, and the
microbiome-and-CAR-T cohort already described above.
The trial is still the highest evidence grade the thread reaches this month.
What it is not is half of the month.
Interception acquires two clinical-grade anchors and changes state.
36990610 (Lancet Oncol) is a
prospective cohort substituting endoscopic surveillance for prophylactic total
gastrectomy in hereditary diffuse gastric cancer — the precursor treated as
something to watch and act on rather than pre-emptively resect.
37046089 (Nature) puts
extrachromosomal DNA amplification inside Barrett's oesophagus, giving the
precursor a molecular event instead of a histological grade. Together they move
interception from too-early to forming. Neither reports an intervention
outcome, and the status change is as much a fact about the thread's small card
count as about the month.
On the meeting in this month. Section 6 shows 6,360 AACR 2023 abstracts,
because the meeting ran 14–19 April. The journal papers above were published
alongside that meeting, not after it. A journal title cannot evidence a
podium-to-paper interval in either direction — it carries no submission date, no
presentation history, and no way to distinguish a paper that followed a talk from
one that was in press when the talk was given. Nothing in this briefing should be
read as the meeting leading the literature or the literature trailing the meeting,
and no clinical-practice card was written for this month for exactly that
reason. Simultaneity has been mistaken for lead in this corpus once already and
it cost a published conclusion.
4. Refutations¶
(a) Landing on a tracked thread¶
37020027 (Nature): a blinded,
randomized trial of sonographers versus AI for cardiac function assessment. This
is the strongest clinical result for machine learning in the month, and the model
that produces it is a single-task echocardiography model. foundation-models
claims that scaling data and parameters is what carries a model into clinical
use; the clinical win here is carried by a narrow model instead. The trial
design is the reason this counts — blinded and randomized is a grade almost
nothing in the AI literature reaches.
Corrected 2026-09-04
This paragraph ended "and the thread moves too-early → contested on it".
That was true of a curve computed from titles. Read from abstracts,
foundation-models is already crisis in February and contested in
March, and this month it moves contested → splitting — the re-read added
the supporting side that a title-only pass had not collected. The reading of
the card is unaffected; what was wrong is the transition it was credited
with causing. Status is derived from the cards, so a card can keep its
meaning and lose its consequence.
36822206 (Immunity): RORγ+
regulatory T cells are the emissaries that carry a microbial regeneration signal
to distant tissue. suppressive-population-inversion tracks cases where a
population nominated for depletion turns out to be doing necessary work, and this
is one: the cost of removing these cells is tissue repair. The thread's Q1 run —
CD200+ CTLs, IL-5-producing CD4 cells with eosinophils, CD39 marking
tumour-reactive rather than exhausted T cells — continues rather than pauses.
37072350 (JITC): a multicentre EBMT registry documents severe cytopenia after CD19 CAR-T, read alongside 37045474 on the endotheliopathy and hemostatic imbalance underlying CAR-T toxicities. The limiting problem in the platform's best-established indication is haematological, not antigen choice — which is a constraint on generalizing the platform, not on choosing targets for it.
(b) Method credibility¶
37130797 (Genome Res): STARR-seq reproducibility is challenged. This matters beyond the assay, because STARR-seq output is used as ground truth by regulatory-genomics models — an unreliable assay does not merely produce bad enhancer calls, it produces confidently wrong benchmarks. Read with 37072822, the ENCODE Imputation Challenge's critical assessment of cross-cell-type epigenomic imputation, and 36973549 on the reliability and accuracy of single-molecule FRET: three audits of established measurement in one month.
37097356 (JAMA): thirty-two randomized trials re-run as nonrandomized database emulations, with the agreement reported as a measured quantity. The observational literature is being tested against its own gold standard rather than trusted, and the escalation this thread tracks — from the model, to the pipeline, to the instrument — here reaches the study design itself.
5. Open items¶
Nothing carried forward.
6. Meetings¶
6,365 abstracts from 2 meetings.
| Meeting | Abstracts | With full text |
|---|---|---|
| AACR Annual Meeting 2023 | 6,360 | 6,360 |
| ASCO Annual Meeting 2023 | 5 | 5 |
7. Regulatory¶
| Event | Count |
|---|---|
| trial-first-posted | 474 (of which 430 genuinely oncology) |
| ema-authorisation | 4 |
| fda-approval | 3 |
3 FDA actions — dated from each notice's own prose, not the table's posting date, which disagrees on 7 of 190 rows and once by enough to move an approval into the wrong month:
- 2023-04-03 — FDA grants accelerated approval to enfortumab vedotin-ejfv with pembrolizumab for locally advanced or metastatic urothelial carcinoma
- 2023-04-17 — FDA approves omidubicel to reduce time to neutrophil recovery and infection in patients with hematologic malignancies
- 2023-04-19 — FDA approves polatuzumab vedotin-piiq for previously untreated diffuse large B-cell lymphoma, not otherwise specified, and high-grade B-cell lymphoma
1 EMA oncology authorisation:
- 2023-04-19 — Akeega
8. Trade press¶
No items. The press archive covers 2024-01 to 2026-08; feeds that cannot be paginated backwards contribute only from the first harvest forward.