June 2026 — Monthly Briefing¶

1. What was read¶

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

A further 317 records (24% of the 1,308 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
529	266	91	388	34

Volume swing against the median month: bioinfo +43%, cancer +43%, sysbio +42%

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 136, Nature 102, Nucleic Acids Res 100, Science 90, Bioinformatics 80, Blood 78.

2. Net change on the thread curve¶

Thread	This month	Cumulative	Status
`agentic-ai`	+4	+36 / −9	forming
`cell-therapy`	+6	+238 / −75	forming
`ctdna-mrd`	+2	+85 / −29	forming
`foundation-models`	+5 / −5	+167 / −118	splitting
`immunotherapy-resistance`	+10 / −1	+268 / −96	forming
`interception`	+5	+162 / −27	forming
`method-credibility`	+1 / −21	+23 / −572	crisis
`neoantigen-vaccines`	+2	+86 / −50	splitting
`neuro-immune-microbiome-axis`	+5 / −1	+227 / −24	forming
`next-gen-checkpoints`	+2 / −2	+78 / −61	splitting
`single-cell-3d-genome`	+3	+92 / −23	forming
`suppressive-population-inversion`	−1	+56 / −106	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¶

Two benchmark metrics were declared invalid in the same journal in the same month. 40739072 (Nat Biotechnol) finds that the silhouette-based metrics underpinning single-cell integration benchmarks are invalid, undermining published method rankings; 40854979 finds that perturbation-prediction benchmarks are inflated by systematic non-perturbation variation, so reported gains over trivial baselines are not real. Together with April's 40500472, three cards in three months say the same thing about different subfields: the leaderboard, not the method, is what has been broken.

Mouse models were told they are not the human tumour. 42156893 (Nat Immunol) states that mouse tumour models do not assemble the same immune microenvironment as human tumours. This is the single card that bears on the largest volume of work in this corpus. The templated preclinical block — nominate a target, reverse suppression, synergise with anti-PD-1 in mice — runs at roughly twenty papers a month across the oncology journals, and this paper says its readout system does not reproduce the thing it is modelling. It does not make that literature wrong; it makes the step from that literature to a trial the unvalidated one, which is exactly where next-gen-checkpoints keeps failing.

A widely repeated survival claim was attributed to selection bias. 42330421 (Cancer Discov) reports that the survival benefit of incidental COVID mRNA vaccination around checkpoint therapy is selection bias, not immune synergy. The original observation entered this corpus as a support in October 2025 (41125896, mRNA vaccines sensitizing tumours to checkpoint blockade). Eight months from striking result to attributed artifact is the fastest full cycle the corpus has recorded, and it is the reason the stance field exists. What it is not is a card against neoantigen vaccines. The exposure is non-tumour-specific mRNA vaccination given alongside checkpoint therapy, and it bears on neither the size nor the predictability of the neoantigen target space, which is what neoantigen-vaccines claims. It was counting as a refutation of a thread it does not address, and it is now neutral there.

Cell therapy reached ten years. 42341302 (NEJM) follows 38 patients a median of 10.1 years after a single tisagenlecleucel infusion. No relapse occurred beyond 5.4 years, and ten-year lymphoma-free survival was 32% in large B-cell lymphoma and 47% in follicular lymphoma. The late-toxicity column is on the same page: a second primary cancer in 9 patients (21% cumulative incidence at ten years), 18% non-relapse mortality, and persistent B-cell aplasia in 44% of long-term responders. Every durability and late-toxicity argument in this thread now has a number to attach to, and the two sets of numbers have to be quoted together.

4. Refutations¶

(a) Landing on a tracked thread¶

Corrected 2026-08-27

This subsection opened by reporting that PRESERVE-003 (41896648) showed a next-generation pH-sensitive anti-CTLA-4 failing to outperform chemotherapy after PD-1 failure, and read it as "the third time in the corpus that a checkpoint target survives a failure by being reformatted and then fails again in the new format". The trial reports the opposite. Its title is "Gotistobart or docetaxel" — trial, comparator, population, stage, and no result — and a failure was supplied for it. Stage 1 reports an overall-survival hazard ratio of 0.46. The reformat-and-fail argument has no instance here and is withdrawn; the card is now supports, and the month's next-generation checkpoint cards no longer sit in this subsection.

What the month actually landed on next-gen-checkpoints is three positive readouts, kept here with their sign corrected rather than moved. 41896648 (Nat Med) reports stage 1 of PRESERVE-003: gotistobart, a pH-sensitive anti-CTLA-4 that selectively depletes regulatory T cells inside the tumour, against docetaxel in metastatic squamous NSCLC that had progressed on both PD-(L)1 blockade and platinum chemotherapy. At 14.5 months' median follow-up, median overall survival was not reached with gotistobart against 10.0 months with docetaxel (hazard ratio 0.46, 95% CI 0.25-0.84, nominal two-sided P = 0.0102), with grade 3 or higher treatment-related adverse events in 42% against 49%. Read it for what it is: 87 patients, stage 1 of two, a nominal P value and a dose-confirmation objective. But it is the engineering fix — pH sensitivity, meant to widen CTLA-4's therapeutic window — producing a result in the population where the existing generation has already failed, which is the scenario this thread was built to watch for.

Alongside it, 42284198 (Blood) brings PD-1/TIGIT-knockdown CD19 CAR-T to the clinic, merging the checkpoint and cell-therapy threads: if the receptor cannot be blocked systemically, delete it in the cell you are already engineering. And 42218899 (Lancet) reports the interim overall-survival analysis of HARMONi-6, in which the PD-1xVEGF bispecific ivonescimab plus chemotherapy beat the PD-1 antibody tislelizumab plus chemotherapy in advanced squamous NSCLC — 27.9 against 23.7 months, hazard ratio 0.66 (95% CI 0.50-0.87), at the cost of more grade 3 or higher treatment-related adverse events and more haemorrhage. That is the head-to-head comparator design against an active checkpoint antibody that the earlier anti-TIGIT programmes never got, and it is the design that makes a next-generation claim checkable.

(b) Method credibility¶

42086853 (Nat Genet): age structure distinguishes selection from causation, and recurrence-based driver calling systematically misassigns drivers. Recurrence is how essentially every cancer driver list in use was built.

42062553 (Nat Methods): imaging spatial transcriptomics carries substantial per-gene false positives and transcript misassignment between neighbouring cells. This is the fourth spatial-transcriptomics credibility card in six months, after January's differential-expression false discovery rate, February's segmentation-error dominance and April's version-dependence. Spatial transcriptomics is simultaneously the most productive method area in this corpus and the one with the most evidence against its own base operations.

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.

143 abstracts from 2 meetings.

Meeting	Abstracts	With full text
European Bioconductor Conference 2026	103	98
AHLI Conference on Health, Inference and Learning 2026	40	40

7. Regulatory¶

Event	Count
trial-first-posted	732 (of which 647 genuinely oncology)
fda-approval	5
ema-authorisation	5

5 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:

2026-06-12 — FDA approves capivasertib with abiraterone and prednisone for PTEN-deficient androgen pathway modulation-naïve or -sensitive prostate cancer
2026-06-12 — FDA approves belzutifan with pembrolizumab for adjuvant treatment of renal cell carcinoma
2026-06-24 — FDA approves sacituzumab govitecan-hziy as monotherapy and in combination with pembrolizumab for first-line treatment of triple-negative breast cancer
2026-06-24 — FDA approves palbociclib with trastuzumab, with or without pertuzumab, and endocrine therapy for the maintenance treatment of HR-positive, HER2-positive metastatic breast cancer
2026-06-30 — FDA approves allogeneic regulatory T cell-based immunotherapy with HSPC and T cells-vldq for use in matched donor hematopoietic stem cell transplantation for adults with hematologic malignancies

1 EMA oncology authorisation:

2026-06-19 — Palbociclib Viatris

8. Trade press¶

53 on-topic of 323 items across 3 feeds (n-stat 308, n-aacr-blog 11, 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.

2026-06-29 — Editors’ Picks, June 2026: Preventing Colorectal Cancer, Targeting Neural Cross-talk to Stop Glioblastoma Recurrence, and More — n-aacr-blog
2026-06-29 — STAT+: FDA’s top gene therapy regulator is leaving his role — n-stat
2026-06-29 — STAT+: Talawar bets bispecific approach can shake up eczema market — n-stat
2026-06-29 — STAT+: Pharmalittle: We’re reading about a broader $35 insulin cap, a former FDA cancer drug chief, and more — n-stat
2026-06-26 — At the AACR Annual Meeting 2026, Cancer Grand Challenges Teams Explore How Cancer Forms and Evolves — n-aacr-blog
2026-06-26 — Supreme Court rules in weedkiller cancer case — n-stat
2026-06-26 — Opinion: Joseph Fraumeni Jr., pioneering cancer genetics researcher, devoted his life to families like mine — n-stat
2026-06-25 — The Long Road to Receiving Care—Examining Cancer Disparities in Rural Areas — n-aacr-blog
2026-06-25 — STAT+: Next-gen CRISPR tools improve editing accuracy in embryos, but also stoke ethical concerns — n-stat
2026-06-24 — Clinical trial set to test two drugs for fast-growing Ebola outbreak — n-stat
2026-06-24 — STAT+: Wagering on FDA changes, Regenxbio will submit Duchenne gene therapy for approval — n-stat
2026-06-24 — Neutrophils not so neutral in tumors after all — n-acir
… and 41 more on-topic items.