January 2026 — Monthly Briefing

1. What was read

1,211 research articles from 33 of the 33 journals on the roster.

A further 294 records (20% of the 1,505 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
551	206	79	639	30

Volume swing against the median month: bioinfo +136%, sysbio +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). Every journal on the roster contributed this month. 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: Nucleic Acids Res 374, Nature 125, JAMA 119, Science 104, Bioinformatics 72, Lancet 56.

2. Net change on the thread curve

Thread	This month	Cumulative	Status
agentic-ai	+1 / −2	+18 / −6	forming
cell-therapy	+4 / −3	+206 / −67	forming
ctdna-mrd	+1	+75 / −24	forming
foundation-models	+7 / −5	+139 / −89	splitting
immunotherapy-resistance	+4 / −2	+224 / −87	forming
interception	+5	+132 / −23	forming
method-credibility	+2 / −11	+20 / −495	crisis
neoantigen-vaccines	+1 / −3	+81 / −44	splitting
neuro-immune-microbiome-axis	+5	+196 / −20	forming
next-gen-checkpoints	+2 / −2	+70 / −49	splitting
single-cell-3d-genome	+1 / −1	+78 / −18	forming
suppressive-population-inversion	−3	+53 / −98	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

Checkpoint blockade was questioned at its own premise, not at a target. 41299179 (Nature) reports that the inhibitory PD-1 axis maintains high-avidity stem-like CD8 T cells: prolonged antigen engagement in tumour-draining lymph node niches sustains TCF-1+PD-1+SLAMF6-high precursors, PD-1 fine-tunes the T-cell receptor input that lets the highest-affinity clones expand, and blockade disrupts that tuning, pushing the most avid precursors to terminal differentiation or death. The authors' own closing question is whether anti-PD-1 buys a short-term effect at the cost of the pool responses draw on — which makes this an argument about how much PD-1 blockade and for how long, not about which receptor to block instead. It is carded neutral on next-gen-checkpoints for exactly that reason: a result about the PD-1 axis cannot count against a claim about targets beyond PD-1 and CTLA-4, and if it leans either way it is a motive for looking beyond them.

Anti-TIGIT's two readouts this month both point the same way, and both are in the oesophagus. SKYSCRAPER-08 is the anti-TIGIT phase 3 that reads out positive in this corpus, in oesophageal squamous carcinoma; in the same month MORPHEUS-EC raised confirmed objective response to 67.7% with tiragolumab added to atezolizumab and chemotherapy, against 53.8% for atezolizumab and chemotherapy and 47.8% for chemotherapy alone in first-line oesophageal cancer, and the investigators read the 14-point margin as additional benefit from the anti-TIGIT antibody. MORPHEUS-EC is phase 1b/2 with an objective-response endpoint and no power for survival, so it is a signal and not a verdict. A class that fails in several tumours and produces two concordant positive readouts in one is a class with an unsolved selection problem rather than a dead target.

Foundation models were audited from four directions at once, and one of them won. Fifty TCR-epitope predictors across twenty-one datasets all fail on unseen epitopes; a single-hidden-layer contrastive model beats deep integration architectures; deep expression models only match linear baselines even after personal-genome fine-tuning and fail entirely on unseen genes; and four leading LLMs queried on 8,432 curated TF-target interactions reach 48-50% balanced accuracy under single-turn prompting — near chance, although multi-turn prompting and the exclusion of pairs with unknown regulation type carry some conditions towards 70%, which the authors read as usable support for regulatory biology. Against that, AlphaGenome beats the best external models in 25 of 26 variant-effect evaluations. The pattern to hold onto is that the wins are on sequence-to-function tasks with hard ground truth, and the failures are on tasks where the evaluation set was constructed by the field being evaluated.

Corrected 2026-08-27

The paragraph above counted a fifth audit — AlphaFold-class predictors failing on flexible and disordered interfaces — and cited it to 41570803. That record is a Cell Systems preview of two other papers, and it reports the opposite: modelling protein dynamics improves binding-site prediction for flexible systems. The failure was read out of the subtitle, "why dynamics and disorder remain a challenge"; the piece itself presents no failure of AlphaFold-class predictors and no data of its own. The card is now neutral and the count in this paragraph is four.

Cell therapy moved off the manufacturing line. Universal base-edited CAR7 T cells give off-the-shelf allogeneic clinical data in T-ALL, and pTalpha potentiates solid-tumour CAR-T through translation-level engineering rather than receptor redesign.

4. Refutations

(a) Landing on a tracked thread

41611245 (JITC) is a post-hoc dissection of why ICOS agonism did not translate in INDUCE-1. It matters more than a single failed programme: ICOS is the target whose first-in-human trial appeared in August 2025 as the one piece of clinical checkpoint activity in a preclinical desert, and it belonged to the wave before the roster being nominated at the time. The clinic testing the previous fashion while the literature nominates the next one is the mechanism that produces this thread's divergence, and this card is its post-mortem.

41482545 (Nat Immunol): which neoantigens are seen depends on efferocytic transfer of dead-cell antigen, not on prediction. 41506790 (JITC): neoepitope prediction algorithms do not recapitulate clinically observed T-cell responses. Two independent statements in one month that the prediction layer under personalized vaccines is not measuring the thing it is used to select on.

41106379 and 41401810 both land on suppressive-population-inversion — TET2-mutant clonal haematopoiesis improving checkpoint therapy, and opposing functions for distinct Treg subsets in colorectal cancer.

(b) Method credibility

41193809 (Nature): specificity, length and luck drive gene rankings in association studies, with GWAS and rare-variant burden tests prioritizing different genes for reasons unrelated to the trait. 41519776 (Genome Biol): in standard spatial differential expression, false discoveries greatly outnumber true findings. Spatial transcriptomics is the modality this corpus records the most method papers in, and this is the first card saying the base test is wrong rather than suboptimal.

5. Open items

Nothing carried forward.

6. Meetings

2 meetings with a written vault this month. These are read and summarised rather than harvested as abstracts, so they do not appear in the counts below.

• 44th J.P. Morgan Healthcare Conference
• Pacific Symposium on Biocomputing 2026

100 abstracts from 1 meeting.

Meeting	Abstracts	With full text
AAAI-40 (AAAI-26) -- biomedical subset	100	100

7. Regulatory

Event	Count
trial-first-posted	758 (of which 646 genuinely oncology)
ema-authorisation	11
fda-approval	1

1 FDA action — 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:

• 2026-01-27 — FDA approves daratumumab and hyaluronidase-fihj with bortezomib, lenalidomide, and dexamethasone for newly diagnosed multiple myeloma

1 EMA oncology authorisation:

• 2026-01-09 — Inluriyo

8. Trade press

34 on-topic of 299 items across 3 feeds (n-stat 287, n-aacr-blog 8, n-acir 4). 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.

• 2026-01-31 — Opinion: STAT readers on shared decision-making, prostate cancer screening, and more — n-stat
• 2026-01-29 — Editors’ Picks, January 2026: Naked Mole Rat’s Cancer Resistance, the Impact of Tissue and Liquid Biopsy Concordance, and More — n-aacr-blog
• 2026-01-29 — STAT+: ImmunityBio’s Soon-Shiong mischaracterized talks with FDA over cancer drug, sources say — n-stat
• 2026-01-28 — STAT+: DeepMind releases AlphaGenome source code, aiming to improve its AI predictions of gene regulation — n-stat
• 2026-01-28 — STAT+: FDA halts gene therapy trials after brain tumor found in patient — n-stat
• 2026-01-28 — STAT+: Startup Altido Therapeutics pursues CAR-T for glioblastoma tumors — n-stat
• 2026-01-28 — Opinion: STAT+: Medicare may start covering multi-cancer early detection tests. It should proceed with caution — n-stat
• 2026-01-28 — Breaking bad macrophages through remodeling with armored CAR-T — n-acir
• 2026-01-27 — Opinion: Sidelining Black women in clinical trials is not just a moral failure. It’s bad science — n-stat
• 2026-01-26 — STAT+: Pharmalittle: We’re reading about pharma lobbying, an FDA gene-editing pathway, and more — n-stat
• 2026-01-23 — Cancer Today’s Winter 2025-2026 Issue: Delaying Stem Cell Transplants in Multiple Myeloma, Emerging Treatments for Brain Metastases, and More — n-aacr-blog
• 2026-01-23 — STAT+: Pharmalittle: We’re reading about drug shortages, llama antibodies for developing meds, and more — n-stat
• … and 22 more on-topic items.