May 2023 — Monthly Briefing¶
1. What was read¶
741 research articles from 33 of the 33 journals on the roster.
A further 217 records (23% of the 958 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 |
|---|---|---|---|---|
| 419 | 178 | 77 | 256 | 28 |
Largest contributors: Nature 103, Nucleic Acids Res 88, JAMA 87, Science 81, Bioinformatics 67, J Clin Oncol 51.
2. Net change on the thread curve¶
| Thread | This month | Cumulative | Status |
|---|---|---|---|
cell-therapy |
+6 / −3 | +37 / −13 | forming |
ctdna-mrd |
+2 / −1 | +13 / −9 | splitting |
foundation-models |
+1 | +9 / −8 | splitting |
immunotherapy-resistance |
+14 | +54 / −13 | forming |
interception |
+6 | +33 / −3 | forming |
method-credibility |
−18 | +0 / −68 | crisis |
neoantigen-vaccines |
+3 / −1 | +15 / −6 | forming |
neuro-immune-microbiome-axis |
+6 | +32 / −4 | forming |
next-gen-checkpoints |
+2 / −1 | +11 / −6 | splitting |
single-cell-3d-genome |
+1 | +10 / −5 | forming splitting → forming |
suppressive-population-inversion |
+4 | +8 / −13 | contested crisis → contested |
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¶
Read first from titles and journals, then re-read from abstracts on 2026-09-04; 43% of this month's cards — the highest share in the first half of 2023 — state something no title carries. Passages below that defer to what "a title will bear" were written under the earlier reading and are left standing where the underlying judgement did not change.
A ctDNA assay makes the treatment decision, in a randomized trial.
36863484 (Ann Oncol): the APPLE
phase 2 assigns osimertinib on plasma T790M monitoring rather than on imaging
progression. ctdna-mrd claims ctDNA is treatment-guiding and not only
prognostic, and this is the rare design that tests that claim directly — the
randomization is over what the assay is allowed to decide. Most ctDNA papers in
this quarter, including this month's, report association with outcome instead,
which is the other half of the thread and does not move it.
A microbiome resistance mechanism becomes a receptor. 37138075 (Nature): microbiome-related immunotherapy resistance is traced to the PD-L2–RGMb pair and reversed by blocking it. This is the step that separates a mechanism from an association — a named ligand and receptor, a blockade, and a reversal — and it lands on two threads at once, since RGMb is also a non-PD-1 blockade target with a stated route to the clinic.
The neural arm measured in patients, with an outcome attached. 37138086 (Nature): glioblastoma remodels human neural circuits and the extent of remodelling tracks survival. Alongside it, 37169842 (Nat Cancer) reports GAP43-dependent mitochondrial transfer from astrocytes into glioblastoma cells. Two distinct mechanisms of tumour–neuron coupling in one month, one of them in humans with survival as the endpoint.
A third source of antigen that is not the host genome.
37198490 (Nature): microbially
derived peptides activate tumour-infiltrating lymphocytes in glioblastoma — a
tumour with almost no mutational neoantigens, which is where the
neoantigen-vaccines scale claim is weakest. Set against April's transposable
element convergence, the target space is being enlarged from three unrelated
directions in eight weeks: transposons, chimeric transcripts, and microbes.
Single-cell chromatin profiling goes multifactorial as a class, not as an
instrument. 36316484 reads several
histone marks simultaneously in single cells;
36344841 profiles chromatin–protein
crosstalk at locus resolution in the same journal and month, and the journal
flagged the pair itself with 37138033.
Three more follow in June (36536148,
36536150,
36593403). single-cell-3d-genome
claims these modalities become routine inputs that downstream pipelines must
ingest; five methods in two months from independent groups is what the front edge
of that looks like, and it is also a warning that the pipelines have nothing
standard to ingest them with yet.
Section 6 shows 26 AACR 2023 abstracts landing in May — stragglers from the April meeting, not a May event. As in April: nothing here should be read as the meeting leading or trailing this literature, because a journal title cannot evidence that interval in either direction.
4. Refutations¶
(a) Landing on a tracked thread¶
37195941 (NEJM): a phase 2 of
peresolimab, a PD-1 agonist, in rheumatoid arthritis. next-gen-checkpoints
claims that targets beyond PD-1 and CTLA-4 deliver a second generation of
checkpoint blockade. The checkpoint-axis drug that reaches a top general
journal with a phase 2 readout this month is neither beyond PD-1 nor blockade —
it runs the axis backwards, into autoimmunity. That is not a failed target; it is
the category moving sideways out from under the claim. Nothing else in the month
contradicts it — the month's other checkpoint entries
(36888930, CD47 with azacitidine in
higher-risk MDS) are carded as support — but the strongest clinical readout in
the axis is pointed elsewhere.
36493337 (JCO): ctDNA is already
changing management in HPV-related oropharyngeal cancer in the absence of
guidelines to support it. This is a refutation about practice rather than
biology, and it sits in the same month as APPLE above: the assay is being acted
on faster than the evidence that acting on it helps is being generated. Both
belong to ctdna-mrd; with 36913694
(Blood), where ctDNA clearance after two cycles predicts progression-free
survival in untreated classic Hodgkin lymphoma, the thread reads +2/−1 this
month — and that third card is prognosis again, which is the half of the thread
that does not move it.
Adjacent and carded as supports on immunotherapy-resistance rather than as a
refutation, but worth reading here:
37205768 (Sci Immunol) reports that
PD-L1 blockade itself activates regulatory T cells in a way that underlies
therapy resistance. The therapy generating its own failure mode is support for
the resistance thread and a constraint on the blockade programme at the same time.
(b) Method credibility¶
Eighteen this month, which is why the thread reads −18 and stays in crisis. Four of them reach furthest, and those four are described below.
Corrected 2026-09-04
This read "Four this month, which is why the thread reads −4". The month is eighteen cards. The fourteen not described are of a piece with the four that are — null models for genomic hypothesis testing, single-cell transformations and pseudo-bulk variance, splicing prediction, CRISPR-screen reproducibility, and how gradient-based interpretations of deep networks for genomics have to be corrected — so the month's shape is unchanged and only its size was wrong. It was wrong because the thread was carded from titles, and an audit usually names its object in the title and its verdict in the abstract.
36930853 (JCO): response rate, progression-free survival and overall survival are argued to have come apart from one another. Almost every clinical readout this corpus records is reported on response rate or progression-free survival, so this is an audit of the units the oncology literature is denominated in.
37024653 (Nat Methods, flagged by the journal with 37024652): in vivo CRISPR off-target detection needed a sensitivity upgrade to see breaks it had been missing. Published off-target profiles are therefore lower bounds set by the assay, not counts of what the editor did — and June extends the same point to prime editors (37156841).
37388906 (Cell Genom): genotype imputation reference panel accuracy is evaluated in sub-Saharan African populations and is a function of who is in the panel. A shared resource whose error rate depends on the ancestry of the person it is applied to, and which is almost never quoted alongside the results built on it.
37165242 (Nature): a draft human pangenome reference, with 37316739 in Nat Biotechnol the following month. The single linear reference underneath essentially every result in this corpus is being replaced rather than extended — and 37217946 shows on a bovine super-pangenome that the graph construction method itself changes which variation is representable. The foundation is not merely incomplete; the choice of replacement is itself a variable that analyses will inherit silently.
5. Open items¶
Nothing carried forward.
6. Meetings¶
26 abstracts from 1 meeting.
| Meeting | Abstracts | With full text |
|---|---|---|
| AACR Annual Meeting 2023 | 26 | 26 |
7. Regulatory¶
| Event | Count |
|---|---|
| trial-first-posted | 532 (of which 471 genuinely oncology) |
| ema-authorisation | 11 |
| fda-approval | 2 |
2 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:
- 2023-05-19 — FDA grants accelerated approval to epcoritamab-bysp for relapsed or refractory diffuse large B-cell lymphoma and high-grade B-cell lymphoma
- 2023-05-31 — FDA approves olaparib with abiraterone and prednisone (or prednisolone) for BRCA-mutated metastatic castration-resistant prostate cancer
1 EMA oncology authorisation:
- 2023-05-04 — Tibsovo
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.