February 2023 — Monthly Briefing¶
1. What was read¶
681 research articles from 32 of the 33 journals on the roster.
A further 174 records (20% of the 855 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 |
|---|---|---|---|---|
| 376 | 187 | 80 | 196 | 16 |
Volume swing against the median month: bioinfo -31%, sysbio -33%
Check the source before reading anything into this. Two benign causes account for most of these. bioinfo every January carries the Nucleic Acids Research database issue and runs high. And several journals simply deposit nothing in some months: across the 44 months read, 4 of 33 journals are absent from at least one, worst being Nat Mach Intell (20 months missing, 81 papers in total) and Bioinformatics (4 months missing, 2,905 papers in total). This month is missing Nat Mach Intell. Queried directly, PubMed returns zero for those journal-months too — the holes are in the source, reproduced exactly by re-running the harvest. So a swing here is not evidence that a field went quiet, and is not usually evidence that the pipeline broke either.
Largest contributors: Nature 100, Science 81, Nucleic Acids Res 73, JAMA 53, J Clin Oncol 47, Bioinformatics 43.
2. Net change on the thread curve¶
| Thread | This month | Cumulative | Status |
|---|---|---|---|
cell-therapy |
+6 / −3 | +12 / −6 | forming |
ctdna-mrd |
+2 / −1 | +5 / −5 | splitting contested → splitting |
foundation-models |
+1 / −2 | +2 / −4 | crisis too-early → crisis |
immunotherapy-resistance |
+16 / −3 | +22 / −5 | forming |
interception |
+5 / −2 | +11 / −2 | forming |
method-credibility |
−11 | +0 / −20 | crisis |
neoantigen-vaccines |
+4 / −1 | +7 / −2 | forming too-early → forming |
neuro-immune-microbiome-axis |
+4 / −2 | +9 / −3 | forming |
next-gen-checkpoints |
+3 / −1 | +4 / −4 | splitting too-early → splitting |
single-cell-3d-genome |
+3 / −1 | +5 / −3 | splitting too-early → splitting |
suppressive-population-inversion |
−3 | +0 / −7 | crisis too-early → 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¶
Written in arrears, and re-read from abstracts
2023 was read after 2024, 2025 and 2026, which still caps what it can support. It was originally carded from titles and journal names alone; on 2026-09-04 every research abstract of the month was re-scanned with the instrument 2026 was read with. 37% of this month's cards state something no title carries, and they were added by that second pass, not the first.
The one trial that tests the ctDNA claim directly will not tell a title-only
reader what it found. 36423745 (Ann Oncol) reports the c-TRAK TN trial,
which used ctDNA mutation tracking to detect residual disease and trigger
intervention in early-stage triple-negative breast cancer. That is the
ctdna-mrd claim built into a protocol rather than inferred from a survival
curve, and it is the only such design in the quarter. The title states the design
and the existence of results and stops there; the card is neutral, and that is a
limit of this month's reading rather than a property of the trial.
A second trial answers the same question from an interaction rather than from a
protocol. 36130300 (Blood) tracks residual disease through two years of
maintenance in 1,280 myeloma patients across TOURMALINE-MM3 and -MM4. Ixazomib
maintenance improved progression-free survival in the patients who were
MRD-positive at randomisation — median 18.8 versus 11.6 months, HR 0.65 — and did
nothing for those who were MRD-negative. That is an MRD-by-treatment interaction:
the assay is not describing risk, it is separating the patients in whom the drug
works from the patients in whom it does not, which is the ctdna-mrd claim. The
paper's title says "for improved prognostication" and the interaction sits in the
body of the abstract, which is why a title-only reading filed this one as a
refutation; see the note in section 4.
Recurrent neoantigens from splice junctions between exons and transposable elements, twice in one issue. 36735774 and 36735776 (Sci Immunol) report the same antigen class from two angles — immunogenic in lung cancer patients, and epigenetically controlled. The load-bearing word is recurrent: these are shared across patients, which enlarges the target space while removing the per-patient manufacturing that the thread's own claim assumes is required.
Engineered cell therapy generalises on two axes at once. 36690811 (Nat Med) reports interim phase 1 results for an allogeneic, off-the-shelf CAR against BCMA in relapsed and refractory myeloma. A non-CD19 target and non-autologous manufacture in the same product is a harder test of the thread than either would be alone.
Spatial single-cell tumour atlases arrive back to back in one journal issue.
36725934 and 36725935 (Nature): the lung tumour immune
microenvironment, and primary and metastatic brain tumours. Two atlases side by
side is spatial multi-omics being handled as an expected deliverable rather than
as a demonstration, which is the modality half of what single-cell-3d-genome
claims.
The neuro-immune-microbiome thread turns forming on two mechanisms, not on
volume. 36693372 (Immunity): dopamine reprogrammes ILC2 metabolism and
shuts down the inflammation those cells drive — a neurotransmitter as the direct
control input, which is the thread's shared-currency clause stated as one
experiment. 36640762 (Cell): immunity directed at the microbiota is what
permits sensory neurons to regenerate, joining all three arms in a single causal
chain.
4. Refutations¶
(a) Landing on a tracked thread¶
CD39 is read two incompatible ways inside eight weeks. 36543958
(Nat Immunol) makes CD39 the suppressive effector of exhausted T cells under
hypoxia — the target-nomination reading, and the one an inhibitor programme needs.
36792124 (JITC), the same month, uses CD39 with CD103 to identify
neoantigen-specific cytotoxic T cells in low-mutation colorectal cancer, and
36574773 (Immunity) in January made it the marker of the tumour-reactive
compartment that predicts checkpoint response. Three papers, two months, one
molecule, and the disagreement is not about effect size but about which side of
the synapse it belongs to. This strains next-gen-checkpoints rather than
supporting it: the thread claims that targets beyond PD-1/CTLA-4 deliver a second
generation of blockade, and a nomination whose sign is unsettled in its own
literature is not yet a target.
Neutrophils invert, with two opposite prescriptions in the same month. 36706760 (Cancer Cell) proposes neutrophil-activating therapy for cancer. 36800412 (Sci Immunol), the same month, reports lung mesenchymal cells reprogramming neutrophils into an immunosuppressive state that promotes breast cancer metastasis. Both are defensible readings of the same cell and they prescribe recruitment and removal respectively. That is a refutation of the depletion logic itself, not a disagreement about one target.
suppressive-population-inversion therefore sits at 0/−4, and the number needs
its denominator before anyone reads a collapse into it. The thread's own
denominator field records that its supporting side is a genre rather than a set
of citable findings — the templated preclinical block that nominates a suppressive
population, deletes it, and reverses immune suppression in mice — running at
roughly a dozen to thirty papers a month across the oncology journals and
producing nothing worth a card. A card is written when a population inverts. So
an all-refutes column is what the reading rule produces by design, and the honest
statement of it is documented inversions against a steady stream of uncarded
nominations.
36658434 (Nat Genet): NOTCH1 mutations drive clonal expansion in normal
oesophageal epithelium but impair tumour growth, with 36702998 reporting
APOBEC mutagenesis as a common process in normal small intestine. interception
needs the measurable precursor and the thing worth intercepting to be the same
population; here they are explicitly not, and the expanded clone may be
protective.
Corrected 2026-08-27
This paragraph read four MRD reports as one block that "stops at prognosis",
and the block was assembled from titles. Re-read against the abstract,
36130300 is not in it: TOURMALINE-MM3/-MM4 reports an
MRD-by-treatment interaction, the abstract opens "MRD evaluation may help to
guide treatment duration", and the card is now supports. The paper is
described in section 3. The generalisation across the other three reports was
never checked against anything but their titles, and it is withdrawn with it.
36084320 (Blood) reports MRD detection in CLL below 10⁻⁴,
36240445 (Blood) sequencing-based risk stratification in childhood ALL,
and 36315929 (JCO) the prognostic value of residual FLT3-ITD in AML.
None of the three is carded. A title that names a prognostic endpoint does not
say whether a treatment decision followed it, and that question is what
ctdna-mrd turns on; what this month can say about the thread is in section 3.
Three more land quietly. 36725933 (Nature) improves immunomodulatory antibody potency by lowering affinity for the target, putting the gain in the antibody's format rather than in the choice of receptor. 36702947 (Nat Med) explains durable antitumour immunity by persistent tumour mutation burden, returning the determinant of response to inside the cancer cell. 36203013 (Nat Biotechnol) reports that post-translational modification reshapes the presented MHC-I repertoire — the antigens a tumour actually displays are not recoverable from the genome that every prediction pipeline takes as input. And 36327160 (Blood) reports interventions and outcomes for myeloma patients needing salvage after a BCMA-directed CAR: the platform reaches the new indication, and a literature on what to do when it stops working is the shape of the bound on that.
(b) Method credibility¶
method-credibility crosses into crisis at 0/−5 this month, which is an
artefact of scoring a one-sided ledger by ratio and not a signal about the field.
The objects are what matter.
36726140 (Genome Biol): the collateral activity of RfxCas13d induces lethality in a knock-in mouse. A widely adopted RNA-targeting effector kills the animal it is expressed in, which means knockdown phenotypes obtained with it are not separable from the tool's own toxicity.
36657438 (Cell Syst): genotyping of microbial communities degrades as
reference genome collections grow. The strain-level calls the microbial arm of
neuro-immune-microbiome-axis rests on move with the reference rather than with
the biology — and the audit and the thread it undercuts are in the same month.
36639584 (Nat Methods) is carded on foundation-models but belongs
here as well: scaled structure prediction returns one structure per sequence, and
36424442 exists in the same month to add back the ligands and cofactors
it omits. The capability and its stated ceiling are co-present, not sequential.
5. Open items¶
Nothing carried forward.
6. Meetings¶
1,541 abstracts from 1 meeting.
| Meeting | Abstracts | With full text |
|---|---|---|
| ASCO Annual Meeting 2023 | 1,541 | 1,541 |
7. Regulatory¶
| Event | Count |
|---|---|
| trial-first-posted | 471 (of which 425 genuinely oncology) |
| ema-authorisation | 4 |
| 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:
- 2023-02-03 — FDA approves sacituzumab govitecan-hziy for HR-positive breast cancer
- 2023-02-09 — FDA grants regular approval to dostarlimab-gxly for dMMR endometrial cancer
2 EMA oncology authorisations:
- 2023-02-20 — Imjudo
- 2023-02-20 — Tremelimumab AstraZeneca
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.