March 2024 — Monthly Briefing

1. What was read

744 research articles from 32 of the 33 journals on the roster.

A further 232 records (24% of the 976 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
442	161	89	262	22

Largest contributors: Bioinformatics 113, Nature 98, Science 93, JAMA 80, Lancet 59, Nucleic Acids Res 46.

2. Net change on the thread curve

Thread	This month	Cumulative	Status
agentic-ai	—	+4 / −0	too-early
cell-therapy	+2	+96 / −30	forming
ctdna-mrd	+2	+32 / −13	forming
foundation-models	+2 / −2	+38 / −27	splitting
immunotherapy-resistance	—	+101 / −35	forming
interception	+1	+62 / −9	forming
method-credibility	−4	+3 / −176	crisis
neoantigen-vaccines	—	+42 / −17	forming
neuro-immune-microbiome-axis	+4	+68 / −8	forming
next-gen-checkpoints	+2 / −2	+29 / −21	splitting
single-cell-3d-genome	—	+26 / −7	forming
suppressive-population-inversion	−1	+23 / −31	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

Simple beats complex, in print, in March 2024. 38428432 (Cell Syst): convolutions are competitive with transformers for protein sequence pretraining. This corpus has dated the simple-baselines-beat-deep-models finding twice — first to 2026, then, after the 2025 backfill, to June 2025. It is here, in a tracked journal, in March 2024. The correction is now the fourth of its kind and the pattern is no longer about any individual claim.

One laboratory ships two pathology foundation models back to back. 38504018 is UNI, a general-purpose self-supervised model pretrained on more than 100 million tiles from over 100,000 diagnostic slides across 20 tissue types and evaluated on 34 tasks. 38504017 is CONCH, a visual-language model trained on 1.17 million image-caption pairs and state of the art across 14 benchmarks. Both are in the same issue of Nature Medicine, both are real, and both card supports — two distinct models with distinct evaluations. What they are not is two groups: they share a senior author and most of their author list. Read them as one group's release of a pair of models, which is a claim about that group's capability and about how far the framing has travelled in imaging, and not as replication.

Corrected 2026-08-27

This paragraph argued that "independent convergence in the same issue is the strongest evidence grade available from literature alone". The convergence was not independent. The inference behind it — same journal, same month, therefore different laboratories — is the one this corpus reaches for most often, and it fails silently whenever a laboratory publishes back-to-back, because nothing in either title reveals the shared authorship. The stances are unchanged; the pair is one piece of evidence, not two.

A neurotransmitter as the shared currency, made explicit. 38489352 (Sci Immunol): gut bacteria-derived serotonin promotes immune tolerance in early life. The neuro-immune-microbiome-axis claim asserts that neurotransmitters are the currency between the neural, microbial and lymphoid arms. This is that assertion as a title, twenty months before the corpus opened the thread.

4. Refutations

(a) Landing on a tracked thread

38200243 (Nat Cancer): SIRPα on tumour-associated myeloid cells restrains antitumour immunity independent of its interaction with CD47. The don't-eat-me axis that the entire CD47 clinical programme was built on is not the operative mechanism. In December 2025 this corpus reads two Nature Immunology papers relocating CD47's function as a "mechanistic post-mortem" arriving five months after the phase 3 failures. The post-mortem was published twenty-one months before them.

38418879 (Nature): anti-TIGIT antibody improves PD-L1 blockade through myeloid and Treg cells — TIGIT's mechanism reassigned away from the T-cell-intrinsic model its clinical programme assumed, two years before SKYSCRAPER-01 reads out negative in July 2026.

38096462 (JCO): four additional doses of PEG-L-asparaginase during consolidation do not improve outcome and increase toxicity in high-risk ALL, randomised — no difference in end-of-consolidation MRD or five-year event-free survival, and adverse reactions in 22.2% against 8.9%. A clean negative escalation trial, and it lands on no tracked thread. It was carded on clinical-practice, whose claim is that meetings lead journals; this is a journal readout that says nothing about a meeting, and all eight of that thread's cards were the same mistake — scoring whether the clinical advice went up or down instead of scoring the thread's actual claim. The card keeps its stance and has no thread.

(b) Method credibility

38416482 (JAMA): blind spots, shortcuts and automation bias in AI clinical models. Read with 38504018 and 38504017 above — the enthusiasm and its audit are again in the same month, not separated by the interval this corpus once recorded.

5. Open items

Nothing carried forward.

6. Meetings

6,642 abstracts from 1 meeting.

Meeting	Abstracts	With full text
AACR Annual Meeting 2024	6,642	6,642

7. Regulatory

Event	Count
trial-first-posted	541 (of which 470 genuinely oncology)
fda-approval	7
ema-authorisation	3

7 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:

• 2024-03-01 — FDA approves amivantamab-vmjw for EGFR exon 20 insertion-mutated non-small cell lung cancer indications
• 2024-03-06 — FDA approves nivolumab in combination with cisplatin and gemcitabine for unresectable or metastatic urothelial carcinoma
• 2024-03-06 — FDA approves inotuzumab ozogamicin for pediatric patients with acute lymphoblastic leukemia
• 2024-03-07 — FDA grants accelerated approval to zanubrutinib for relapsed or refractory follicular lymphoma
• 2024-03-19 — FDA grants accelerated approval to ponatinib with chemotherapy for newly diagnosed Philadelphia chromosome-positive acute lymphoblastic leukemia
• 2024-03-21 — FDA approves safety labeling changes regarding DPD deficiency for fluorouracil injection products
• 2024-03-22 — FDA approves mirvetuximab soravtansine-gynx for FRα positive, platinum-resistant epithelial ovarian, fallopian tube, or primary peritoneal cancer

8. Trade press

36 on-topic of 267 items across 3 feeds (n-stat 253, n-aacr-blog 11, n-acir 3). 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.

• 2024-03-28 — STAT+: Pharmalittle: We’re reading about a GLP-1 side effect, U.K. clinical trial transparency, and more — n-stat
• 2024-03-28 — STAT+: U.K. becomes the first country to release detailed data on sponsors that fail to register clinical trials — n-stat
• 2024-03-27 — STAT+: Borrowing an idea from cancer immunotherapy, scientists make old-mice immune systems youthful again — n-stat
• 2024-03-27 — STAT+: Brazil’s plan for low-cost CAR-Ts — n-stat
• 2024-03-27 — Stem-immunity hubs predict response to anti-PD-1 in NSCLC — n-acir
• 2024-03-26 — STAT+: Viking Therapeutics to move its oral obesity drug into Phase 2 study after early success — n-stat
• 2024-03-21 — Dosing, Surgery, and Treatment Resistance—Seeking New Options for Kidney Cancer — n-aacr-blog
• 2024-03-21 — STAT+: Getting CRISPR treatments to patients remains a major challenge. Could lab-developed tests help? — n-stat
• 2024-03-20 — STAT+: Immuno-oncology startup launches with $150 million, ex-Surface, Bluebird exec as CEO — n-stat
• 2024-03-20 — STAT+: New gene therapy, to be priced at $4.25 million, has already transformed children’s lives — n-stat
• 2024-03-20 — STAT+: A scientific mystery highlights the blind spots in genomics databases — n-stat
• 2024-03-19 — Women in Cancer Research: 25 Years of Progress — n-aacr-blog
• … and 24 more on-topic items.