October 2025 — Monthly Briefing¶
1. What was read¶
910 research articles from 32 of the 33 journals on the roster.
A further 307 records (25% of the 1,217 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 |
|---|---|---|---|---|
| 587 | 194 | 94 | 314 | 28 |
Volume swing against the median month: general +25%
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: JAMA 141, Nature 119, Science 107, Nucleic Acids Res 78, Bioinformatics 60, Cell 59.
2. Net change on the thread curve¶
| Thread | This month | Cumulative | Status |
|---|---|---|---|
agentic-ai |
+2 / −1 | +16 / −3 | forming |
cell-therapy |
+2 | +196 / −63 | forming |
ctdna-mrd |
— | +71 / −23 | forming |
foundation-models |
+9 / −2 | +124 / −79 | splitting |
immunotherapy-resistance |
+4 / −4 | +218 / −81 | forming |
interception |
+2 | +124 / −23 | forming |
method-credibility |
−15 | +18 / −459 | crisis |
neoantigen-vaccines |
−2 | +77 / −40 | splitting forming → splitting |
neuro-immune-microbiome-axis |
+14 | +184 / −19 | forming |
next-gen-checkpoints |
−2 | +68 / −42 | splitting |
single-cell-3d-genome |
+5 | +71 / −17 | forming |
suppressive-population-inversion |
−4 | +51 / −85 | 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¶
October is the emptiest checkpoint month in the corpus. A full scan of the eight oncology journals for TIGIT, LAG-3, TIM-3, CD47, SIRPα, VISTA, BTLA, NKG2A, B7-H3, CD73, HPK1, PCSK9 and galectin returns zero hits. No target from July, August or September survives. What October nominates instead is ligand-side and innate machinery — CMTM6 as the PD-L1 stability chaperone, the sialic acid/Siglec axis on stroma — with no clinical programme behind any of it. The one axis with genuine momentum is CD40 agonism (40816292). The field has moved from blocking new inhibitory receptors to agonising an old stimulatory one.
Two independent immunology papers put exhaustion in the stress response rather than the receptor. 40930086 attributes CD8 dysfunction to oxidative-stress-induced telomere instability; 41005293 attributes it to chronic ATF4 activation under intratumoral hypoxia. Same issue of Immunity, same phenotype, two different upstream causes, neither of them an inhibitory receptor. If exhaustion is a genome-damage and proteostasis problem, checkpoint blockade is treating a symptom.
The neuro-immune axis produced its cleanest single title. 41166460 (Science) encodes enteric nervous system responses to both microbiota and type 2 inflammation with regional resolution — all three arms of the thread in one experiment. Fourteen cards opened on this thread in October and every one of them is a support, which is the second-largest month the thread has had, behind 2026-07's twenty. And 41138728 moves the axis out of homeostasis: tumours co-opt an inter-organ neuroimmune circuit to escape immune surveillance.
Corrected 2026-09-04
This said "Six cards opened on this thread in October, the highest monthly
count in the corpus." Six was the title-only count; the 2026-09-04 abstract
re-scan puts October at fourteen, matching the +14 in section 2. The
superlative does not survive the re-scan either: 2026-07 opened twenty cards
on neuro-immune-microbiome-axis, eighteen of them supports. What is
unusual about October is not the size but the sign: fourteen cards and not
one refutation among them.
Clinical AI: the counterweight stopped degrading without recovering. 40963011 (Nature) is a genuine experimental negative — delegation to AI increases dishonest behaviour — and a JAMA audit examined the demographic evidence behind FDA authorization of AI devices. Neither is a null randomized clinical trial, and no benchmarking-crisis methodological critique returned.
4. Refutations¶
(a) Landing on a tracked thread¶
41094201 (Nat Immunol): TIGIT, the
receptor classified as co-inhibitory and pursued as a blockade target, promotes
tissue-protective functions in T cells. This is the strongest single card on
suppressive-population-inversion and it lands on a target with active phase 3
programmes.
40935830: damage-induced IL-18 stimulates thymic NK cells that limit regeneration — an activating signal on a beneficial-by-assumption effector producing the harm. 41092896 is an Immunity preview of Ewing-Crystal et al. in Nature, a paper this corpus does not hold: injury fibroblasts occupy dynamic states that limit the initial injury and promote repair, so they are not purely scar-forming. The reading survives; the record carrying it is a commentary on another journal's work, which the earlier version of this line did not say and the card did not record. 40700578: MDSCs are administered as a GVHD therapy rather than depleted as a target.
41028524 (Nat Genet): genetic effects on disease susceptibility barely overlap effects on survival. Every polygenic score built for risk and reused for prognosis inherits this.
(b) Method credibility¶
41099710: sites with no complementarity to the microRNA seed bind and repress as well as the top canonical sites do, so seed pairing is not required. Seed complementarity is the rule underneath essentially every miRNA target-prediction algorithm in use. But the paper bounds its own damage, and an earlier version of this entry — which called it the month's most consequential card — left the bound out: these 3'-only sites are estimated at under 1% of endogenous target sites. The rule is broken in principle and rarely in practice, which is a narrower claim on the prediction literature than "the rule does not hold".
Two independent reference-genome results — choose your human genome reference wisely and reference bias under chromosomal reorganization — plus ParaMask asserting major uncorrected multicopy-region biases in whole-genome sequencing, and synthetic associations created by epistasis, which means a mapped GWAS signal need not mark a causal variant.
On foundation models, one capability boundary, and it is a sharp one. 40789967 runs vision-language models against MaCBench and separates what they do from what they are credited with: near-perfect at equipment identification and standardised data extraction, failing at spatial reasoning, cross-modal synthesis and multi-step logical inference. That is a direct limit on the perception-to-reasoning step every agentic system depends on.
Corrected 2026-08-27
This entry counted three boundaries in one month and leaned hardest on the one it called first. The other two are not independent evidence.
40931098, "Vision language models excel at perception but struggles with scientific reasoning", is the Nature Computational Science research briefing accompanying MaCBench in the same journal in the same month, with no abstract of its own. Same finding, restated by its own publisher.
41023433, "What does a language model know about proteins?", is a Nature Methods front-matter piece. PubMed carries no abstract, and the title states no result — the card read an interrogative as a finding against the thread. It is now neutral.
Three capability boundaries was one benchmark, its accompanying briefing, and a question.
40818455 (Cell): RNA Pol II inhibition kills cells independently of transcriptional loss — a whole drug class's mechanism reinterpreted.
5. Open items¶
Nothing carried forward.
6. Meetings¶
No abstract book published this month. The abstract layer runs 2023-02 to 2026-08 and is seasonal by nature — meetings are annual, so most months are empty here and that is the expected shape, not a gap in the harvest.
7. Regulatory¶
| Event | Count |
|---|---|
| trial-first-posted | 287 (of which 258 genuinely oncology) |
| fda-approval | 4 |
| ema-authorisation | 1 |
4 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:
- 2025-10-02 — FDA approves lurbinectedin in combination with atezolizumab or atezolizumab and hyaluronidase-tqjs for extensive-stage small cell lung cancer
- 2025-10-08 — FDA approves cemiplimab-rwlc for adjuvant treatment of cutaneous squamous cell carcinoma
- 2025-10-23 — FDA approves belantamab mafodotin-blmf for relapsed or refractory multiple myeloma
- 2025-10-24 — FDA approves revumenib for relapsed or refractory acute myeloid leukemia with a susceptible NPM1 mutation
8. Trade press¶
44 on-topic of 326 items across 3 feeds (n-stat 308, n-aacr-blog 12, n-acir 6). Headlines are selected by subject keyword, not recency — most of what these feeds carry is health policy and industry finance the corpus does not read.
- 2025-10-29 — STAT+: Natera, known for spotting cancer recurrence, wades into early detection —
n-stat - 2025-10-29 — CRATERs on the tumor cell surface are hubs for antitumor immunity —
n-acir - 2025-10-28 — STAT+: Miriam Merad’s 20-year quest to understand one cell is starting to rewrite the rules of cancer immunotherapy —
n-stat - 2025-10-27 — Risky Business: Breast Cancer Risk Factors and What We Know About Them —
n-aacr-blog - 2025-10-27 — STAT+: Hope and ‘no regrets’: How pancreatic cancer surgeon Dr. Michael Zinner faced a pancreatic cancer diagnosis —
n-stat - 2025-10-27 — STAT+: For Intellia Therapeutics, CRISPR drug’s safety concern is an existential threat —
n-stat - 2025-10-27 — STAT+: Syndax leukemia drug wins FDA expanded approval —
n-stat - 2025-10-27 — STAT+: Intellia pauses two late-stage CRISPR gene-editing trials after emergence of safety concern —
n-stat - 2025-10-27 — STAT+: CAR-T therapies from Cabaletta, Bristol Myers Squibb induce remissions in patients with autoimmune disorders —
n-stat - 2025-10-24 — Editors’ Picks, October 2025: Drug Delivery Technology, Anticancer Effects of Alzheimer’s Protein, and More —
n-aacr-blog - 2025-10-23 — Frontiers in Cancer Science 2025: Attend the Singapore International Cancer Conference —
n-aacr-blog - 2025-10-23 — STAT+: FDA delivers split decision on GSK’s blood cancer drug Blenrep, approving its use in some cases —
n-stat - … and 32 more on-topic items.