The trial in question — known as Exenatide-PD3 — was a multicenter, double-blind, randomized, placebo-controlled study conducted across six hospitals in the United Kingdom. Led by researchers at University College London and funded by the National Institute for Health and Care Research, the trial enrolled 194 participants with Parkinson’s disease who received either subcutaneous once-weekly exenatide or placebo over 96 weeks. The primary endpoint was change in the MDS-UPDRS Part III motor score, assessed in the off-dopaminergic-medication state at 96 weeks.

When results were published in The Lancet in February 2025 (DOI: 10.1016/S0140-6736(24)02808-3), the trial did not meet its primary endpoint: no statistically significant difference was found between the exenatide and placebo arms at 96 weeks. A secondary observation — a modest motor improvement in the exenatide group at the 60-week timepoint — drew attention but was not sufficient to support a disease-modifying conclusion.

The EOC signals that The Lancet’s editors have concerns about aspects of the paper and that an investigation is underway. The notice does not constitute a retraction. The specific nature of the editors’ concerns, the scope of the investigation, and its timeline are not disclosed in the EOC itself — which is standard practice at this stage of a journal inquiry.

The stakes are notable. Parkinson’s disease remains without any approved disease-modifying therapy. The hypothesis that GLP-1 receptor agonists might confer neuroprotection has drawn substantial scientific and commercial interest, and the Exenatide-PD3 study was the first Phase 3 test of that hypothesis in a neurodegenerative disease. An EOC does not change clinical guidance and should not alter patient or prescriber decisions; it is a signal that editors are reviewing evidence — not a finding of misconduct or error.

The investigation’s outcome is pending. The Vital Record will update this story as further information becomes available.

Researchers led by Masaki Terabe at the National Cancer Institute tested a heterologous prime-boost strategy: a first shot of ChAdOx1, a simian adenovirus, followed by a booster with modified vaccinia Ankara (MVA). The vector pair targets two tumor antigens, P1A and Gpr149, in orthotopic implants of the SB28 glioblastoma line — a model chosen precisely because it resists immune checkpoint blockade (ICB).

The regimen produced therapeutic regression of established tumors in the mouse brains. When the team examined what was doing the work, the dominant population was CD103+CD69+CD8+ tissue-resident memory (TRM) T cells — cells that, once seeded in a tissue, can mount local responses without relying on fresh recruits from the bloodstream.

In a key transfer experiment, glioblastoma-derived antigen-specific TRM cells taken from vaccinated mice and injected intracranially into naïve animals were sufficient on their own to protect those recipients from a subsequent tumor challenge. That result suggests the cells carry the protective capacity, not just bystander activity.

Notably, layering ICB on top of vaccination added no benefit — an observation the authors interpret as evidence that the bottleneck in GBM immunity is T-cell priming, not checkpoint suppression.

The ChAdOx1 adenoviral vector component is already clinical-stage: it is the same platform used in AstraZeneca’s Oxford–AstraZeneca COVID-19 vaccine — which uses ChAdOx1 alone; MVA is not part of that product. The shared vector ancestry lowers one regulatory hurdle for ChAdOx1-based approaches, though the full ChAdOx1/MVA prime-boost combination used here is a separate construct. The antigens used here are murine, human GBM antigen targets remain to be defined and validated, and no human trial is registered or planned. The findings are preclinical and have not been peer reviewed.

Correction, 2026-06-22: An earlier version of this article stated that the “ChAdOx1/MVA backbone” underpins AstraZeneca’s Oxford–AstraZeneca COVID-19 vaccine. That is incorrect: the AstraZeneca COVID-19 vaccine (AZD1222/Vaxzevria) uses ChAdOx1 as a single vector; MVA is not part of it. The relevant prior clinical experience is with ChAdOx1 alone, not the ChAdOx1/MVA prime-boost combination studied here. The sentence has been corrected.

This is preclinical research conducted in rodents. No human clinical trial has been registered, initiated, or planned based on these findings.

What the Researchers Did

GMP cells are bone marrow progenitors that give rise to multiple myeloid effector populations, including macrophages, dendritic cells, and granulocytes. The USC team used gene-editing techniques to introduce CAR constructs into GMP cells — creating what they term CAR-GMPs — and evaluated their activity in transplanted mouse tumor models.

Unlike CAR-T cells, which target tumors through T-cell cytotoxicity, CAR-GMP cells act via phagocytosis and innate immune signaling. The research group proposed that myeloid-lineage effectors may be better positioned to penetrate immunosuppressive tumor microenvironments that resist T-cell infiltration — a hypothesis supported by the mouse model results but untested in humans.

In CD19-positive leukemia models, anti-CD19 CAR-GMP cells significantly reduced tumor burden compared with controls. In HER2-positive solid tumor models, anti-HER2 CAR-GMPs showed measurable anti-tumor activity. The team also tested the approach in a mouse model of chronic granulomatous disease, a primary immunodeficiency caused by defective phagocyte function, and observed restoration of some phagocytic activity.

What the Study Does Not Establish

Mouse tumor models are valuable for hypothesis generation but are unreliable predictors of human response. Key unknowns include durability of CAR-GMP activity in vivo, cytokine release risk, manufacturing scalability for clinical-grade production, and whether the anti-tumor signal in rodents translates to human cancer biology.

Mouse models frequently fail to predict human responses — including in oncology, where numerous promising preclinical programs have not replicated clinically. The distance between these results and a clinical proof of concept is substantial.

USC Keck School of Medicine. CAR-GMP progenitor cell therapy. Cell. 2026. DOI: 10.1016/j.cell.2026.05.043. PMID: 42320470.

IMbrave050 (NCT04102098) enrolled 668 patients with high-risk HCC, defined by pathological features including microvascular invasion, satellite nodules, or large tumor size, and randomized them one-to-one to atezolizumab 1,200 mg plus bevacizumab 15 mg/kg IV every three weeks for 12 months versus active surveillance. The study was the first large randomized Phase 3 trial of an immune checkpoint combination in the adjuvant HCC setting.

The first interim analysis, published in The Lancet in 2023 (Qin et al., Lancet 2023;402:1835–1847; DOI: 10.1016/S0140-6736(23)01796-8), reported a recurrence-free survival benefit for the combination arm, with OS data too early for analysis.

The second interim analysis, now published by Yopp and colleagues in the Journal of Hepatology (2026;84:1102–1111), shows an updated RFS hazard ratio of 0.90 (95% CI 0.72–1.12) — a point estimate still favoring the combination but no longer statistically significant. More critically, overall survival data remain immature: the OS hazard ratio at second interim is 1.26 (95% CI 0.85–1.87). That wide confidence interval does not exclude meaningful harm, does not confirm benefit, and reflects the relatively short follow-up relative to the natural history of resected HCC.

The numerical OS point estimate exceeding 1.0 — meaning more deaths in the treatment arm than in active surveillance at this interim — warrants caution in extrapolating the RFS finding to a net benefit conclusion. Mature OS remains the regulatory gold standard in oncology adjuvant trials; the current RFS estimate, no longer statistically significant, is insufficient alone to establish a favorable benefit-risk profile. Additional follow-up and OS maturity are required before the adjuvant role of atezolizumab-bevacizumab in HCC can be fully evaluated.

Yopp A, et al. IMbrave050 second interim. J Hepatol. 2026;84:1102–1111. Qin S, et al. Lancet. 2023;402:1835–1847. DOI: 10.1016/S0140-6736(23)01796-8.

The study, led by Peter Sasieni (Wolfson Institute of Population Health, Queen Mary University of London) and Alejandra Falcaro, tracked cervical cancer mortality in English women by birth cohort according to the timing and coverage of the NHS HPV vaccination program launched in 2008. Among the cohort with vaccine coverage of 88–90% — those who received the jab through the school-based program at ages 12–13 — no cervical cancer deaths were recorded during the five-year observation window (women aged 20–24 in 2020–24). The authors report an 80% reduction in cervical cancer mortality in women aged 20–24 in 2015–19 (a period capturing the first routinely vaccinated cohorts as well as catch-up recipients), and a 69% reduction among women aged 25–29 in 2020–24, each compared with modelled expected mortality in the absence of vaccination.

What the evidence shows — and what it does not

The zero-mortality finding in the highest-coverage cohort represents a landmark in the epidemiology of vaccine-preventable cancer. Prior evidence from Scotland and Sweden had demonstrated near-elimination of HPV-16/18-associated cervical intraepithelial neoplasia (CIN2+) in vaccinated cohorts, and multiple countries reported falling incidence rates. England’s national cohort study, spanning the NHS cervical screening programme and the ONS death registry, now adds mortality as an endpoint — the hardest clinical signal in cancer epidemiology.

The authors are careful to note that the zero-death finding applies to the highest-coverage birth cohort at the ages studied (early-to-mid thirties), during a five-year window. The absence of deaths does not constitute proof of elimination across all ages or all HPV-associated cancers; follow-up will be required as the cohort ages into the decade of peak cervical cancer incidence. The comparison with unvaccinated historical controls also carries the assumptions inherent in such designs.

Cervical cancer caused by HPV types 16 and 18 — the two strains targeted by all licensed HPV vaccines, and by the bivalent Cervarix vaccine initially offered in England — accounts for approximately 70% of cervical cancer burden. Types not targeted by the original bivalent vaccine (31, 33, 45, 52, 58) account for a further 20%. England began offering the nonavalent Gardasil 9 (covering nine HPV types) in 2021, which should broaden coverage going forward.

England’s HPV vaccination programme

The NHS launched the national HPV immunisation programme in September 2008, offering two doses of Cervarix to girls in Year 8 (ages 12–13), with a catch-up programme extended to girls aged up to 18. England switched to Gardasil 4 in 2012, and then to Gardasil 9 in 2021. The programme was extended to boys in 2019. Coverage in the primary cohort (Year 8 girls) reached 80–90% in the early years of the programme and has been sustained above 85% in most years since, making England one of the highest-coverage national programmes in Europe.

The Lancet analysis translates that coverage into a mortality outcome for the first time at national scale.

Global implications

Cervical cancer remains the fourth most common cancer in women worldwide, responsible for approximately 350,000 deaths annually — the large majority in low- and middle-income countries where HPV vaccination coverage and cervical screening remain limited. The England result will reinforce advocacy for high-coverage national programmes in settings where both are available. It also provides a mortality endpoint for the mathematical models used to estimate the benefit-cost ratio of vaccination programmes.

WHO’s 2030 targets for cervical cancer elimination — 90% of girls vaccinated by age 15, 70% of women screened by age 35 and 45, and 90% of women identified with disease treated — require substantial scale-up. The England data add urgency to that agenda.

Sasieni P, Falcaro A, et al. “HPV vaccination and cervical cancer mortality in England: a population cohort study.” The Lancet. 2026. DOI: 10.1016/S0140-6736(26)00918-9. PMID: 42309117.

Correction (2026-06-19): Two errors corrected by post-publication fact-check. (1) Lead author Peter Sasieni’s institution was listed as King’s College London; the Lancet paper (PMID 42309117) lists his affiliation as the Wolfson Institute of Population Health, Queen Mary University of London. (2) The 80% cervical cancer mortality reduction was attributed to “catch-up vaccination cohorts”; the source assigns this figure to women aged 20–24 in 2015–19 — an age-period cell that includes both routine and catch-up vaccinees, not catch-up recipients exclusively. The comparator for both figures is modelled expected mortality in the absence of vaccination, not unvaccinated historical controls per se.

A preprint posted June 16, 2026 on medRxiv — not yet peer reviewed — reports that sleep regularity, defined as the day-to-day consistency of sleep-wake timing, outperformed sleep duration as a predictor of incident disease across a sweep of 199 conditions in a large prospective cohort of UK Biobank adults.

The study in brief

Lead author Daniel P. Windred and colleagues at the Flinders Health and Medical Research Institute (Sleep Health), Flinders University, South Australia, drew on objective accelerometer data from 60,998 UK Biobank participants (mean age 62.8 years, 55% female), totaling roughly 10 million hours of recorded sleep-wake behavior. Over a 9.5-year follow-up, the team compared multivariable-adjusted incident risks for regular vs. irregular sleepers and for short vs. adequate-duration sleepers across all 199 disease and disorder categories.

The headline result: irregular sleep was associated with elevated incident risk for 131 diseases and disorders — more than double the 63 conditions linked to short sleep duration. For 90 of those conditions — spanning circulatory, metabolic, digestive, renal, infectious, neurological, and musculoskeletal disease as well as mental disorders — sleep irregularity was the superior predictor. Short sleep duration was the superior predictor for only 9 conditions.

Perhaps the most clinically suggestive finding for bedside counseling: in the 83% of conditions where sleep duration alone explained some disease risk, adding sleep regularity to the model improved predictive performance.

What this doesn’t tell us

As with all prospective cohort analyses, association is not causation. The authors note that UK Biobank participants are not representative of all populations — the cohort skews older, whiter, and healthier than the general UK population, and generalizability to other groups is uncertain. The study cannot rule out residual confounding, and no intervention was tested.

The preprint has not completed peer review. Key figures — especially the condition-count comparisons — should be treated as preliminary until reviewed and published.

Why it matters despite the caveats

Sleep duration has dominated clinical sleep-health messaging for decades. If these findings survive peer review, they would add weight to a growing literature suggesting that regularity — going to bed and waking at consistent times — deserves equal or greater attention in sleep-health guidance. The authors conclude that sleep regularity “should be considered an essential dimension of sleep health.”

This article is based on a medRxiv preprint (Windred et al., posted June 16, 2026) that has not been peer reviewed. Findings are preliminary.

Led by researchers at Stanford University, the team derived cell-type-specific protein signatures from plasma and trained models to infer how quickly neurons, astrocytes, immune cells, skeletal myocytes, respiratory epithelial cells, and dozens of other populations were accumulating biological wear. The work, by Ding, Bot, Chen and colleagues in the lab of Tony Wyss-Coray, was accompanied by a companion paper in the same issue (DOI: 10.1038/s41591-026-04447-x).

What the clocks found

Across the cohort, 20–25% of individuals showed accelerated aging in a single cell type, while 1–3% showed accelerated aging in ten or more cell types simultaneously. The clocks were then used to predict incident disease and mortality over up to 15 years of follow-up — a time window that allowed the team to observe who developed conditions like Alzheimer’s disease and amyotrophic lateral sclerosis (ALS) after the initial blood draw.

The APOE4 genotype produced one of the study’s sharpest signals. Carriers of two APOE4 alleles — a group already known to carry elevated Alzheimer’s risk — showed older astrocytes but paradoxically younger macrophages compared with APOE3 carriers. Critically, among APOE4 homozygotes, those whose astrocytes showed extreme biological aging faced triple the risk of developing incident Alzheimer’s disease over follow-up, while individuals with youthful astrocytes saw a reduced risk. The APOE2 genotype showed inverse associations.

For ALS, the association was even more pronounced. Individuals with extremely aged skeletal myocytes — the muscle cells that motor neurons innervate — had a 12.7-fold higher risk of developing ALS compared with those whose skeletal myocytes appeared biologically youthful.

Limitations

Several caveats apply. This is an observational, prospective cohort study; the associations reported cannot establish causation. The biological-age clocks measure proteins circulating in blood as proxies for what is happening inside specific cell types — an indirect inference, not a direct tissue measurement. Whether intervening on cell-type aging trajectories would alter disease outcomes remains untested.

Nonetheless, the authors report that a composite polycellular aging risk score stratified mortality risk across multiple cohorts and proteomics platforms, suggesting these signatures may be reproducible. The framework, the authors argue, offers a cellular-resolution map of human physiology that could eventually inform risk stratification — though clinical translation remains a future step.

The JAMA paper, led by the DPP Outcomes Study Group, followed 1,173 participants originally randomly assigned to intensive lifestyle intervention (n=385), metformin (n=385), or placebo (n=403) at 27 sites (NCT00004992, NCT00038727). At 25-year follow-up, lifestyle intervention reduced multimorbidity incidence compared with placebo (hazard ratio [HR], 0.79; 95% CI, 0.68–0.93) whereas metformin had no significant effect (HR, 0.91; 95% CI, 0.78–1.07).

Multimorbidity was defined as two or more of 15 prevalent conditions using the CMS Chronic Condition Data Warehouse framework. The lifestyle arm’s protective effect was most pronounced when restricted to dyads of the highest-burden conditions (HR 0.57; 95% CI, 0.38–0.85), suggesting the deepest benefit accrues against the costliest co-morbidity pairs, consistent with lifestyle modification’s broad metabolic and inflammatory impact.

That metformin failed to show the same benefit is the provocative finding. Metformin effectively reduced diabetes incidence in the original DPP; that protection did not translate into fewer cumulative conditions at 25 years. The authors propose that metformin’s mechanism—primarily reducing hepatic glucose output—does not address the inflammatory and metabolic pathways that drive multimorbidity beyond glucose control.

The implications for clinical practice are substantial. Metformin is among the most widely prescribed agents in the world, partly on the assumption that diabetes prevention confers broad health protection. The DPP 25-year data suggest that assumption needs reexamination. Lifestyle modification, despite its well-documented challenges in sustained implementation, showed durable protection against a composite of chronic disease burden that no medication in the trial could replicate.

Correction, 2026-06-16: An earlier version of this article stated multimorbidity was defined as 2+ of 8 conditions and listed eight diseases by name. The JAMA source defines multimorbidity as 2+ of 15 prevalent conditions (CMS Chronic Condition Data Warehouse). The per-arm sample sizes have also been corrected (lifestyle n=385, metformin n=385, per published data; previously stated as 380 and 390). The HR 0.57 has been reattributed to dyads of the costliest conditions, which is its correct source context.

Scale of the problem

PPH affects an estimated 17 million women following vaginal birth and 10 million following caesarean section each year. Pooled prevalence is 12.6% (95% CI 10.1–15.2%) for vaginal births and 30.9% (95% CrI 24.9–37.6%) for caesarean births. The global economic burden is estimated at $10.4 billion (95% CrI $9.8–13.2 billion) per year, comprising $3.6 billion in direct health-system costs and $6.8 billion in broader societal costs.

The diagnostic blind spot

One of the series’ most striking findings concerns how PPH is detected. A systematic review found that subjective visual estimation of blood loss — the method most commonly used at the bedside — has a pooled sensitivity of just 48% (95% CI 44–53%) at vaginal birth, meaning it misses more than half (52%) of PPH diagnoses. In response, the WHO has updated its clinical guidance. The updated guidelines introduce a new early-action treatment trigger: clinicians should initiate treatment when objectively measured blood loss reaches ≥300 mL plus any abnormal haemodynamic sign (such as tachycardia or hypotension). The established diagnostic thresholds — ≥500 mL for vaginal birth and ≥1,000 mL for caesarean section — are retained in the consolidated guidelines; the new lower threshold is a prompt to intervene before the full diagnostic threshold is reached, not a replacement of it.

Treatment evidence

A 2025 Cochrane systematic review and network meta-analysis (NMA) of 122 randomised controlled trials involving 121,931 women compared uterotonic agents for prevention of PPH (Gallos et al.; Cochrane Database of Systematic Reviews; DOI: 10.1002/14651858.CD011689.pub4) — a separate Cochrane publication cited within the Lancet Prevention paper (Paper 2). Combinations of oxytocin plus misoprostol, or oxytocin plus ergometrine, were the most effective prophylactic regimens. Among single agents, oxytocin and carbetocin were identified as most effective. Ergometrine alone and the oxytocin-plus-ergometrine combination are no longer recommended as routine first-line choices due to an increased risk of hypertension and associated adverse effects.

The three Lancet papers are: Coomarasamy et al., DOI: 10.1016/S0140-6736(26)00902-5; DOI: 10.1016/S0140-6736(26)00903-7; DOI: 10.1016/S0140-6736(26)01031-7. All are peer-reviewed. The uterotonic NMA is a separate Cochrane publication: Gallos et al., Cochrane Database of Systematic Reviews, DOI: 10.1002/14651858.CD011689.pub4 (2025).

Correction — June 15, 2026: Two framing errors have been corrected. (1) The treatment section previously referred to a “Cochrane NMA” without identifying it as a separate 2025 Cochrane publication (Gallos et al., DOI: 10.1002/14651858.CD011689.pub4) distinct from the three Lancet Series papers; the text now attributes it correctly and notes it is cited within Lancet Paper 2. (2) The WHO definition section previously rendered the new ≥300 mL + haemodynamic-sign threshold as part of a replacement “definition” of PPH; in fact it is a new early-action treatment trigger, and the established diagnostic thresholds (≥500 mL for vaginal birth; ≥1,000 mL for caesarean section) are retained in the consolidated guidelines. Both have been corrected in the text above.

PREPRINT NOTICE: This study has NOT been peer reviewed. It is posted on bioRxiv and has not been accepted by a journal. Findings should not be used to guide clinical decisions.

Researchers at the University of Washington have used saturation genome editing (SGE) to measure the functional impact of thousands of variants in two hereditary cancer-risk genes, RAD51D and XRCC2, according to a preprint posted June 13, 2026, on bioRxiv (DOI 10.64898/2026.06.12.731983). The corresponding author is Lea Starita (Department of Genome Sciences, University of Washington).

What is SGE? Saturation genome editing is a laboratory technique that systematically introduces every possible single-nucleotide variant in a target gene into cells and measures the effect on cell fitness — enabling functional classification of variants at scale. This study was conducted entirely in cell lines (preclinical); results have not been validated in patient cohorts.

Scale and accuracy:

• RAD51D: 5,412 variants measured for cellular fitness; fitness scores discriminated pathogenic from benign variants with AUC = 0.994 (near-perfect)
• XRCC2: 3,743 variants measured for cellular fitness; AUC = 1.000 (perfect discrimination in this dataset)
• Additionally, 2,876 RAD51D variants and 2,069 XRCC2 variants were assessed for effects on RNA expression

RNA splicing finding: Integration of RNA data revealed that 24% of RAD51D loss-of-function missense variants act through RNA-mediated mechanisms (primarily aberrant splicing), compared to only 5% in XRCC2 — a mechanistic insight with potential implications for future variant classification.

Why this matters: RAD51D and XRCC2 encode proteins involved in homologous recombination DNA repair. Most variants in these genes are classified as variants of uncertain significance (VUS), limiting clinical utility. This SGE dataset could, after peer review and regulatory consideration, support reclassification of thousands of VUS.

Critical caveats: This is a preprint, not peer reviewed. The study was performed in cell lines, not patients. SGE is a research tool, not a currently available clinical test. Peer review and independent replication are required before these datasets can inform clinical variant classification guidelines.

Correction — June 15, 2026: The original headline stated that SGE “Resolves Nearly All Uncertain Variants,” which overstates what the preprint reports. The preprint asserts the datasets “provide strong, splice-resolved functional evidence to support variant classification” — a contribution to a multi-step curation process, not a standalone resolution of VUS. The headline has been corrected to “Characterizes Nearly All Variants… Supporting Future Clinical Classification.” The body text, which correctly says the data “could, after peer review and regulatory consideration, support reclassification,” was accurate and unchanged.

The GASTRIC-PICU trial enrolled 4,700 children aged 0 to 16 years across 24 pediatric intensive care units — 23 in the United Kingdom and 1 in Switzerland — who were receiving invasive mechanical ventilation and starting enteral feeds. After excluding 240 from the intention-to-treat analysis, 4,460 children were analyzed: 2,352 randomized to the no-routine-assessment group and 2,348 to usual care. The median age was 8 months; 42.6% were female. Recruitment ran from June 2023 through December 2025.

What Each Group Received

Children in the intervention arm had gastric residual volume (GRV) checking removed from their feeding protocol entirely; nurses assessed feed tolerance using clinical signs alone — vomiting, abdominal distension, and similar indicators. Children in the control arm received GRV checks at least every 6 hours, the standard PICU practice, with feeds paused or reduced when volumes were deemed excessive.

Primary Results

The trial had two co-primary endpoints. On the clinical endpoint — a composite of survival and days free from mechanical ventilation at 30 days — the no-GRV group was noninferior to usual care. Both groups achieved a median of 25 ventilator-free days (IQR 21–27). The adjusted odds ratio was 0.95 (95% CI, 0.86–1.05), and the per-protocol analysis was consistent (adjusted OR 1.01; 95% CI, 0.90–1.13).

On the nutritional endpoint, children in the no-GRV group met a mean of 80.3% of their energy requirements by 72 hours, compared with 76.8% in the usual-care group — an adjusted mean difference of 3.2 percentage points (95% CI, 1.3–5.2; P < .001), a statistically significant superiority finding.

The results suggest that routine GRV monitoring does not protect critically ill children from worse ventilator outcomes or reduced survival, while its removal yields a small but significant improvement in early calorie delivery. Whether that nutritional gain translates into longer-term clinical benefit was not established in this trial.

Limitations: The pragmatic design precludes blinding of clinical staff. The study population was drawn from UK and Swiss PICUs, limiting direct generalizability to other settings. The nutritional benefit is a surrogate measure at 72 hours, not a long-term clinical outcome.

This article describes a clinical research finding in critically ill, ventilated children enrolled in a hospital trial. Decisions about feeding protocols in intensive care are made by critical care teams based on individual patient circumstances.

Researchers at Skane University Hospital in Lund, Sweden enrolled 85 adults with major depressive disorder, dysthymia, or bipolar depression who had clinically elevated scores on the Snaith-Hamilton Pleasure Scale (SHAPS), a validated measure of anhedonia. In the single-center, double-blind, placebo-controlled trial, participants were randomly assigned to add flexible-dose oral pramipexole — a D2/D3 dopamine receptor agonist approved for Parkinson’s disease and restless legs syndrome — or placebo on top of their existing treatments for 9 weeks.

The trial met its primary endpoint. Adults with mood disorders who received pramipexole showed a significantly greater reduction in SHAPS scores compared with placebo (mean difference: -4.04; 95% CI: -6.89 to -1.18; p = 0.006; Hedges’ g = 0.62), an effect size in the moderate range. Exploratory neuroimaging analyses found that pramipexole was associated with relative preservation of reward-related ventral striatal activation on fMRI — biological evidence suggesting the drug engaged its intended dopaminergic target. A separate 6-month open-label extension found improvements sustained, though without a placebo comparator this phase cannot establish whether benefit persisted due to the drug alone.

Important limitations: This is a single-center RCT with 82 participants in the primary analysis, making it a preliminary finding that requires replication in larger, multicenter trials. The study population was specifically adults with mood disorders and elevated anhedonia — results cannot be generalized beyond that group. The trial tested pramipexole as an add-on strategy, not as a replacement for standard antidepressant therapy, and pramipexole is not approved for depression or anhedonia. No head-to-head comparison with standard antidepressants was conducted.

This article reports preliminary research findings and does not constitute medical advice. Patients should not change medications without consulting a physician.

What the study found

Researchers examined four mutations commonly found in clonal haematopoiesis — JAK2V617F, TET2 loss of function (LOF), TP53 loss of function, and DNMT3A — across both preclinical mouse models and two human population cohorts.

In atherogenic mice, sleep and exercise did curtail clone expansion in animals carrying Jak2V617F or Tet2-LOF mutations, but not in Trp53-LOF (the mouse equivalent of human TP53-LOF) or Dnmt3a-mutant animals. On a second, more clinically important measure, however, the picture differed: sleep and exercise also lessened atherosclerosis — the cardiovascular complication that makes clonal haematopoiesis dangerous — in Jak2V617F-, Tet2-LOF-, and Trp53-LOF-driven disease, independent of clone size. Dnmt3a-driven clones were the only ones that responded to neither measure.

In two independent human datasets, moderate-to-vigorous physical activity was associated with lower prevalence of non-DNMT3A-driven clonal haematopoiesis. These human findings are observational and cannot establish causation on their own; the mouse experiments provide the mechanistic evidence for a causal effect in those two mutation types.

What this means — and what it does not

The study does not support clinical recommendations at this stage. Anyone who knows or suspects they carry a CHIP mutation should discuss monitoring and lifestyle guidance with their physician, not modify their routines based on this study alone.

The mutation dependency is the study’s most clinically relevant message: DNMT3A mutations are the most common CHIP variant in older adults, and this research suggests they are not modified by sleep and exercise. Patients with DNMT3A-driven CHIP should not assume lifestyle changes will reduce their clonal burden.

The open-label, dose-escalation study enrolled 54 healthy adults. Forty-four completed the two-dose regimen and were included in the primary immunogenicity analysis. After the second dose, 100% (44/44) developed detectable Lassa-specific IgG antibodies, with titres in the range associated with protection in non-human primate challenge models. The vaccine was well-tolerated; no serious adverse events were attributed to the investigational product.

LASSARAB uses a killed recombinant rabies virus as its vector backbone — the same fundamental platform as licensed inactivated rabies vaccines. The Lassa glycoprotein complex (GPC) insert is displayed on the surface of the inactivated particle, intended to elicit both antibody and T-cell responses to the envelope protein that mediates cellular entry. Because the vector is inactivated, LASSARAB is not a live vaccine, a property that simplifies cold-chain management and deployment in resource-limited settings.

Lassa fever affects an estimated 100,000 to 300,000 people per year in West Africa, with a case-fatality rate of approximately 15–20% among hospitalised patients. No licensed vaccine exists. The WHO has listed Lassa fever as a priority pathogen since 2017; the Coalition for Epidemic Preparedness Innovations has funded multiple candidates, of which LASSARAB is one.

Whether the IgG titres observed translate to clinical protection is not yet established. No validated immune correlate of protection exists for Lassa fever — a gap that complicates accelerated development under animal rule pathways. Phase 2 and Phase 3 trials with efficacy endpoints in endemic populations are required before regulatory filing. At least four other Lassa vaccine candidates are in clinical development, including mRNA-based approaches; the LASSARAB result strengthens the overall signal that antibody responses against Lassa GPC are achievable and well-tolerated in humans.

Researchers mapped the functional connectivity profiles of DMG tumours across paediatric patient cohorts, finding that tumours in the thalamus are embedded in thalamocortical relay circuits while tumours in the brainstem — the subtype known as diffuse intrinsic pontine glioma — are wired into networks governing vital autonomic and consciousness functions. In two independent validation cohorts, greater network integration correlated with shorter survival, suggesting that the extent to which a tumour has colonised critical circuitry shapes the clinical course.

The most consequential implication concerns a small subset of patients. Brainstem DMGs cannot be surgically approached without catastrophic damage to life-sustaining functions — a constraint that has defined and limited treatment for decades. Thalamic DMGs occupy a less absolute anatomical boundary. The authors propose that for selected patients with thalamic DMG and lower network integration scores, surgical debulking — not currently standard practice — warrants prospective study. They explicitly do not recommend surgery for the majority of DMG patients on the basis of these findings.

The analysis is retrospective and hypothesis-generating. The study does not establish a causal relationship between network disruption and improved outcomes, and the authors acknowledge that the network-survival correlation may reflect tumour biology rather than a tractable intervention target. Prospective validation is required before any change to clinical management.

Median survival in unselected DMG is 9 to 11 months from diagnosis. ONC201, a dopamine receptor antagonist, received accelerated FDA approval in H3K27M-mutant DMG in 2024 based on response rate data; overall survival benefit remains under evaluation. The Nature analysis adds a structural and network-level dimension to DMG biology that molecular profiling had not previously captured.

This work is a preprint and has not undergone peer review.

The pfKelch13 C580Y mutation disrupts the ring-stage susceptibility of Plasmodium falciparum to artemisinins, slowing parasite clearance after treatment. It was first identified in the Greater Mekong Subregion, where it contributed to artemisinin-based combination therapy failures. Subsequent genomic surveillance detected C580Y in Rwanda in 2020 and Uganda in 2021; its appearance at 79% prevalence in a Zambian province — in the heart of sub-Saharan Africa’s high-transmission zone — represents a qualitatively new stage of the mutation’s westward spread.

Day-3 positivity — the persistence of detectable parasitaemia three days after initiating artemisinin therapy — is the clinical correlate of partial resistance. At 28%, the rate in Western Zambia approaches thresholds recorded in the Mekong during the period when resistance began driving clinical failures. The critical distinction between partial and full resistance is that current ACT partner drugs, including lumefantrine and amodiaquine, remain active against C580Y strains, and cure rates with complete ACT courses remain high. The margin for error narrows as artemisinin clearance slows.

Zambia uses artemether-lumefantrine as its first-line ACT. A 79% mutation prevalence in a high-transmission province would, if confirmed, constitute a Category I resistance situation under WHO definitions — the threshold at which treatment policy review is triggered. The authors state that findings have been submitted to the WHO Global Malaria Programme’s Malaria Threat Registry. Independent replication is required.

The case report, which has not yet been peer reviewed, is the first published histological evidence that base editing can correct a disease-causing point mutation at therapeutically relevant efficiency in a living human organ.

What base editing is — and why it matters

Unlike earlier CRISPR-Cas9 approaches, which create double-strand DNA breaks and rely on the cell’s repair machinery to introduce changes, base editors use a modified, nickase form of Cas9 paired with a deaminase enzyme to make precise single-nucleotide changes without breaking both strands of the DNA helix. The approach can convert an A to a G (adenine base editors, or ABEs) or a C to a T (cytosine base editors), making it well suited to correcting the class of point mutations responsible for many hereditary diseases — including the Glu342Lys (E342K) substitution that causes the PI*ZZ allele in SERPINA1.

YOLT-202 delivers its base-editing machinery as lipid nanoparticle-encapsulated mRNA directly to hepatocytes, the liver cells where SERPINA1 is expressed. The approach targets hepatocytes because they are both the site of the pathological protein aggregation that causes liver injury and the source of the circulating AAT protein that, when deficient, leads to pulmonary emphysema in AATD patients.

The biopsy

The biopsy was taken ten weeks after the patient received a single intravenous infusion under the Phase I/Ia dose-escalation trial (NCT07193615). Sequencing of biopsy-derived hepatocyte DNA showed correction of 54% of PI*ZZ alleles by Sanger sequencing and 57% by Illumina deep sequencing. Whole-genome sequencing and targeted amplicon sequencing at predicted off-target sites revealed no detectable off-target edits at the sensitivity levels tested.

The 54–57% correction figure represents on-target allele editing in sampled liver tissue. Whether that percentage translates to a proportional increase in circulating normal AAT protein — and whether such levels would be sufficient to prevent disease progression in lung or liver — will be assessed in subsequent cohorts.

Context and caveats

This is a single-patient case report from a Phase I/Ia trial designed primarily to assess safety and dosing. No safety findings are reported in the preprint beyond the biopsy sequencing data. As an unreviewed preprint, the findings should be interpreted with appropriate caution: this is an early signal, not a clinical result.

Even so, few gene-editing technologies have produced direct molecular confirmation of on-target correction in a human organ at this efficiency. Prior gene therapies for AATD, including gene-augmentation approaches that add a functional SERPINA1 copy without correcting the underlying mutation, have shown clinical benefit but do not eliminate the misfolded protein that accumulates in the liver and drives hepatotoxicity. A base-editing approach that corrects the mutation itself offers the theoretical advantage of restoring normal protein folding at the source.

The candidate, pEVAC-PS, was built by the University of Cambridge and its spin-out DIOSynVax using a computational platform that maps structurally conserved epitopes across the sarbecovirus subgenus — the coronavirus lineage that includes SARS-CoV-2, SARS-CoV-1, and a wide range of bat coronaviruses with zoonotic spillover potential. Rather than target a single strain’s spike sequence, the approach aims for shared vulnerabilities that variants are unlikely to mutate away.

The open-label, dose-escalation trial enrolled 39 healthy adults aged 18–50 at two NIHR facilities in Southampton and Cambridge between December 2021 and September 2023. Participants received two doses of pEVAC-PS delivered intradermally via the PharmaJet Tropis needle-free jet injector across four escalating dose cohorts. The primary endpoint was safety and tolerability; immunogenicity was secondary.

The vaccine was well tolerated at all four dose levels, with no significant safety concerns reported — a critical hurdle for any novel vaccine platform entering humans for the first time.

On immunogenicity, the highest-dose cohorts generated neutralizing antibody responses against SARS-CoV-2 Delta and Omicron variants and showed increased reactivity at RBD regions — including a conserved epitope corresponding to that bound by the broadly neutralizing monoclonal antibody S309 — consistent with the vaccine’s cross-sarbecovirus design goal. The authors note that interpretation was complicated by high pre-existing antibody levels and variable Omicron exposure histories during the trial period.

The researchers describe pEVAC-PS as the first AI- and computer-simulation-designed vaccine to complete human Phase I testing.

Important caveats apply. This was a small Phase I study powered to assess safety and preliminary immunogenicity only. No efficacy data — whether the vaccine prevents infection or disease — exist in humans. The trial is published in the Journal of Infection (2026).

VISTA enrolled 310 adults with obesity or overweight in a randomised, placebo-controlled design. Participants on elecoglipron 75 mg lost 10.5% of body weight by week 26 — versus 0.6% with placebo — and weight loss had not plateaued by week 36, where the gap widened to 11.8% versus 0.3%. These are Phase 2 data; confirmatory Phase 3 trials are needed before regulatory submissions.

The companion SOLSTICE trial (n=404) tested elecoglipron in adults with type 2 diabetes. The 75 mg dose cut HbA1c by 1.9 percentage points from baseline at 26 weeks (placebo: −0.2%), and drove 7.7% weight loss (placebo: −1.7%). Ninety percent of treated participants reached an HbA1c below 7%, and 85% reached 6.5% or lower.

Safety across both trials was consistent with the GLP-1 receptor agonist class: adverse events were predominantly mild-to-moderate gastrointestinal, discontinuations were uncommon, and no liver safety signals or treatment-linked serious hypoglycaemia were identified by investigators.

AstraZeneca is moving directly to Phase 3 with the EMBOLD programme (obesity/overweight) and the ELUMINATE programme (T2D monotherapy and combination with dapagliflozin), plus dedicated cardiovascular and renal outcomes trials.

The announcement arrives weeks after Eli Lilly’s orforglipron (Foundayo) received FDA approval for weight management — the first oral GLP-1 pill cleared for obesity in the U.S. — and as Novo Nordisk’s oral semaglutide pursues its own obesity indication. Elecoglipron’s Phase 2 weight-loss figures are broadly comparable to orforglipron’s Phase 2 results, but head-to-head and Phase 3 data will ultimately define competitive positioning. Elecoglipron is not approved for any indication.

A new SARS-CoV-2 sublineage called BA.3.2.2 — a heavily mutated Omicron descendant — is turning up disproportionately in children compared with adults, according to a preprint from Yunlong Cao’s immunology group at Peking University. Genomic surveillance reviewed in the paper shows BA.3.2.2 is significantly enriched in pediatric populations globally, unlike concurrent variants XFG and NB.1.8.1, which circulate more evenly across age groups.

The mechanism appears rooted in immune imprinting — the way a first SARS-CoV-2 encounter shapes all subsequent antibody responses. Adults who were infected with or vaccinated against the original Wuhan ancestral strain built up antibodies encoded by a germline gene segment called IGHV3-53/66. These broadly cross-reactive antibodies recognise a part of BA.3.2.2’s spike protein that the virus has not yet mutated away. Children born after 2021, however, were first exposed to Omicron subvariants — never to the ancestral strain — and their immune systems instead generated antibodies encoded by different gene segments (IGHV2-5 and IGHV5-51). Those Omicron-specific antibodies neutralise dominant current strains well, but BA.3.2.2 appears to have evolved in a way that evades them almost entirely, resulting in markedly lower neutralising titres against this specific variant in younger children.

A companion preprint from Columbia University (Wu et al.) used pseudovirus neutralisation assays across 36 participants spanning adults, school-age children (ages 3–10), and infants and toddlers (ages 6–28 months) and independently confirmed lower geometric mean titres against BA.3.2.2 specifically in both paediatric cohorts, while titres against NB.1.8.1 and XFG were comparable across all age groups.

The Cao lab paper raises a concern that if BA.3.2.2 sustains transmission in children, it could accumulate further mutations capable of also evading IGHV3-53/66 antibodies, potentially enabling wider cross-age transmission. The authors argue the findings may inform paediatric booster strategy. No specific booster recommendation is made; these are preliminary findings requiring peer review.

Primary source: Niu X. et al., “Lack of ancestral SARS-CoV-2 imprinting promotes BA.3.2.2 infection in children,” bioRxiv 2026. DOI: 10.64898/2026.06.05.730251. Not peer reviewed.

Correction, June 10, 2026: The original title stated “While Adults Stay Protected.” Post-publication fact-check confirmed that the Cao lab preprint limits adult protection to those with prior exposure to the ancestral (Wuhan) SARS-CoV-2 strain or ancestral-strain mRNA vaccines; adults whose immune history is Omicron-only are described as “weakly imprinted” and would not retain the same protection against BA.3.2.2. The title has been updated to reflect this specificity. The body of the article, which correctly qualified adult protection to those with ancestral imprinting, was not changed.

The drug is apitegromab, a fully human monoclonal antibody from Scholar Rock that selectively inhibits the activation of myostatin, a protein that restrains muscle growth. Apitegromab is investigational and not approved for any use; it is available only through clinical trials. In the randomized, double-blind, placebo-controlled EMBRAZE study (NCT06445075), 102 adults with overweight or obesity and without diabetes were randomized 1:1 to tirzepatide plus apitegromab (10 mg/kg) or tirzepatide plus placebo, with every participant receiving tirzepatide throughout, according to the trial registry and the published report.

What the trial found

The primary endpoint was the change from baseline in total lean body mass, measured by DXA scan, at 24 weeks. At week 24, apitegromab produced a least-squares mean of 1.9 kg less lean-mass loss than placebo (80% confidence interval 1.2–2.7; P = 0.001) — “despite similar total body weight loss between groups,” the authors write, a result they frame as 54.9% retention of lean mass relative to placebo (DOI).

Selective targeting of myostatin preserved lean mass when combined with tirzepatide — and, critically, did so without blunting overall weight loss.

That last point is the crux. The goal is not simply more muscle; it is to make weight loss “higher quality” — proportionally more fat, less muscle — while keeping the scale moving. The pharmacodynamics tracked the mechanism: in apitegromab-treated participants, trough drug concentrations and total latent myostatin both rose over time and plateaued at roughly 16 weeks.

On safety, the two arms looked alike. Adverse events occurred in 39 of 51 apitegromab-treated participants (76%; 95% CI 63–86%) versus 36 of 51 placebo-treated participants (71%; 57–81%). Serious adverse events were balanced, affecting one of 51 participants (2%; 0–10%) in each arm. The authors describe apitegromab as “well tolerated.”

The caveats worth keeping

This is a proof-of-concept study, and its limits are real. The trial enrolled 102 people and ran 24 weeks — short relative to how long patients stay on weight-loss therapy, and too brief to show whether preserved DXA-measured lean mass translates into preserved strength, physical function or hard clinical outcomes. The primary endpoint was reported with an 80% confidence interval rather than the conventional 95%, a wider margin for declaring an effect that readers should weigh. Lean body mass by DXA is also a surrogate: it captures tissue quantity, not muscle quality or performance. And the study excluded people with diabetes and capped age at 65, so it says nothing yet about the older, frailer patients in whom muscle loss is most consequential.

EMBRAZE establishes that myostatin inhibition can shift the composition of incretin-driven weight loss in a controlled setting. Whether that shift improves how patients actually feel and function will take larger, longer trials with functional endpoints to answer.

Investigators in Japan and Taiwan enrolled patients who had undergone curative resection of colorectal cancer, completed standard-of-care therapy, and then tested positive for circulating tumor DNA (ctDNA) — molecular evidence of residual disease — while showing no radiological sign of recurrence. Between July 2020 and June 2023, 243 such patients were randomized 1:1 to six months of the oral chemotherapy trifluridine/tipiracil (FTD/TPI; Lonsurf) or to placebo, in a double-blind design embedded in the CIRCULATE-Japan platform.

The primary endpoint was investigator-assessed disease-free survival (DFS). Median DFS was 9.30 months with FTD/TPI versus 5.55 months with placebo — a hazard ratio of 0.79 (95% CI 0.60–1.05, P = 0.107). The confidence interval crosses 1.0 and the result did not reach statistical significance, so the primary endpoint was not met.

A nearly four-month gap in median DFS that the trial could not confirm as real — a reminder that an encouraging point estimate is not a positive trial.

A signal without significance

The numerical separation is the kind of result that invites over-reading. But with only 243 patients, ALTAIR was not powered to call a hazard ratio of 0.79 a win, and its authors do not. Acting on a ctDNA relapse signal with FTD/TPI, they conclude, did not significantly improve DFS in patients without radiological disease.

The toxicity tradeoff was unambiguous. Grade 3 or higher hematologic adverse events affected 73.0% of FTD/TPI-treated patients versus 3.3% of placebo-treated patients, with no new safety signals.

ALTAIR tested a specific bet: that starting an established chemotherapy at the moment of molecular recurrence, rather than waiting for visible disease, would change outcomes. It does not settle whether a more active regimen, or a different drug, might. What it does show is that ctDNA positivity alone is not yet a validated trigger for this intervention — the detection technology is ahead of the treatment evidence. Larger trials testing other agents against molecular residual disease will determine whether the early-warning signal can be converted into benefit.

Investigators sequenced 12,685 trial participants — adults with type 2 diabetes and elevated cardiovascular risk, randomized to dapagliflozin 10 mg or placebo. Among them, 121 carried a pathogenic or likely pathogenic variant in a high-confidence cardiomyopathy gene (76 dilated, 25 hypertrophic, 25 arrhythmogenic). Over a median 4.2 years, dapagliflozin was associated with a hazard ratio of 0.18 (95% CI 0.04–0.86) for hospitalization for heart failure in carriers, versus 0.70 (95% CI 0.57–0.86) in noncarriers, with a P value for interaction of 0.03. The carrier confidence interval is wide — consistent with anything from a large effect to a modest one — because the carrier group is so small.

A wide gap in absolute terms, on small numbers

The absolute risk reduction was 13.0% in carriers and 1.0% in noncarriers (P interaction 0.03). Most carriers — 82% — had no prior heart failure; among those, dapagliflozin reduced absolute risk by 12.8%, against 0.6% in noncarriers (P interaction 0.01).

As a hypothesis to be tested rather than a clinical recommendation, the authors raise the possibility that SGLT2 inhibitor treatment might be started early to prevent heart failure in carriers of cardiomyopathy variants — something they say must first be tested in a dedicated trial.

This is a small, exploratory, hypothesis-generating subgroup analysis: the carrier group is tiny, the confidence interval wide, and an association within a subgroup is not proof of causation. It does not establish that people with type 2 diabetes in general benefit, and it is not a basis for anyone to start dapagliflozin to prevent heart failure — the carrier group is genetically defined and can be identified only by sequencing. Writing in Nature Medicine, the authors say the results “need to be confirmed in a prospective, dedicated trial of preventive HF treatments in carriers.” DECLARE-TIMI 58, sponsored by AstraZeneca, enrolled 17,190 patients and completed in September 2018.

The double-blind, placebo-controlled phase 3a trial randomized 274 adults across 46 centres in six countries (the USA, China, Japan, Serbia, Slovakia, and South Africa). All were on stable once-daily basal insulin, with or without metformin, and entered with a mean HbA1c of 8.8%. Participants were assigned 2:2:1:1 to CagriSema 2.4 mg/2.4 mg (n=90), CagriSema 1.0 mg/1.0 mg (n=93), or pooled dose-matched placebo (n=91) for 40 weeks.

What the trial found

On the primary endpoint — mean HbA1c change from baseline to week 40 — the high-dose arm fell 2.33 percentage points (SE 0.08) versus 0.66 points (SE 0.11) on placebo. The 1.0 mg arm fell 2.10 points. Against placebo, that is an estimated treatment difference of −1.68 percentage points (95% CI −1.95 to −1.41) at the higher dose and −1.44 points (95% CI −1.71 to −1.17) at the lower dose, both p<0.0001.

CagriSema “met the primary endpoint, with statistically significant and clinically relevant HbA1c reductions versus placebo,” the authors write, alongside “robust bodyweight reduction and no additional risk of hypoglycaemia.”

The signature for an insulin add-on is weight: rather than the gain insulin tends to cause, CagriSema produced bodyweight reductions of 10–12%, a key secondary endpoint.

Safety tracked the GLP-1 class. Adverse events — mostly mild-to-moderate gastrointestinal complaints — affected 80% of the high-dose arm (72/90), 71% of the low-dose arm (66/93), and 71% of the placebo arm (65/91). No group had any severe hypoglycemia. One participant in the 1.0 mg arm died of a malignancy that investigators judged unrelated to treatment.

Two caveats temper the readout. At 40 weeks and 274 participants, REIMAGINE 3 is short and small, and the comparator is placebo rather than intensified insulin — so it shows that CagriSema beats doing nothing more, not that it beats the standard escalation. The trial was funded by Novo Nordisk, which employs several of the authors. CagriSema remains investigational and is not approved for type 2 diabetes — or any other use — in the USA, EU, or elsewhere.

A note on regulatory status: orforglipron (brand Foundayo, from Eli Lilly) was FDA-approved on April 1, 2026 for chronic weight management only — in adults with obesity, or with overweight plus a weight-related condition. It is not approved for type 2 diabetes. Lilly has said it plans to file for a diabetes indication on the strength of the ACHIEVE program, but that submission has not been made or cleared. The use studied here remains investigational and is not currently available.

The randomized, double-blind trial enrolled 546 adults whose HbA1c stayed between 7.0% and 10.5% despite insulin glargine, with or without metformin and an SGLT-2 inhibitor, across 72 sites in five countries. Participants took once-daily oral orforglipron at 3 mg (n=137), 12 mg (n=132), or 36 mg (n=136), or placebo (n=141), each alongside titrated glargine. The primary endpoint was the change in HbA1c from a baseline of 8.50% to week 40 for the 12-mg and 36-mg doses; the 3-mg dose was a key secondary endpoint. Of those randomized, 92.9% completed the trial.

The numbers

“At week 40, the mean changes from baseline in HbA1c were −1.58%, −1.88%, and −1.82% with orforglipron, 3 mg, 12 mg, and 36 mg once daily, respectively, vs −0.79% with placebo.”

Because placebo plus titrated glargine alone lowered HbA1c by 0.79 points, the drug’s own contribution is the placebo-subtracted treatment difference: −1.08 points (95% CI, −1.33 to −0.83) at 12 mg and −1.03 points (95% CI, −1.28 to −0.77) at 36 mg, each P<.001. Both higher doses met the primary endpoint and were superior to placebo. The 3-mg dose, a key secondary, produced a difference of −0.78 points (95% CI, −1.02 to −0.55). The authors report that every key secondary outcome — including the share of participants reaching HbA1c targets below 7.0% and at or below 6.5% — favored orforglipron over placebo with statistical significance. Body weight fell 4.8% at 12 mg and 5.4% at 36 mg, while the placebo group gained 0.2%.

Gastrointestinal events, mostly mild to moderate, were the most common adverse effect among orforglipron-treated patients. Treatment discontinuation for any adverse event ranged from 3.6% to 9.6% across the orforglipron doses, versus 3.6% on placebo; discontinuation for GI events specifically ranged from 1.5% to 6.7% across doses, versus 0.7% on placebo. The authors report no increase in clinically significant (level 2) hypoglycemia among orforglipron-treated patients versus placebo. As a 40-week efficacy study, ACHIEVE-5 was not designed to assess cardiovascular outcomes or long-term safety.

The central contrast is mechanistic. According to the abstract, double-strand breaks made by CRISPR-Cas9 in early human embryos “result in frequent aneuploidy and large deletions, revealing a repair deficiency in early human embryos and limiting the clinical application of this technology.” Base editors instead introduce DNA nicks and mismatches. The authors report that editing at PCSK9 and HBG was efficient and, unlike Cas9-induced breaks, “did not result in either chromosomal abnormalities or large deletions.” Small insertions or deletions were rare, and off-target activity depended on the guide RNA. This is an absence-of-detected-harm result in a small, non-peer-reviewed study, not a demonstration that the approach is harm-free.

A note on the targets: the abstract collapses the paralogous fetal-hemoglobin genes HBG1 and HBG2 into a single label, HBG. Secondary coverage in Nature describes three genes edited — PCSK9, HBG1, and HBG2 — so the preprint’s “two targets” and the “three genes” reported elsewhere refer to the same work.

Delivery decided viability

Method mattered. According to the abstract, delivering the editor as protein at fertilization or at the pronuclear stage “allowed normal development to the blastocyst stage and the derivation of edited stem cell lines,” whereas delivering it as RNA “resulted in early embryo arrest.” Edits were mosaic — some cells carried the change, others did not — and Egli told Nature the base editors “can have damaging effects on the embryo,” signaling the approach is not clinically ready.

“These base editors — they can have damaging effects on the embryo. So why would you use it if you don’t fully understand that?” — Dieter Egli, to Nature

Geneticist Fyodor Urnov was sharply critical, telling Nature that the work reads like a how-to manual for “baby improvers” pursuing forays beyond the ethical pale, and noting that IVF combined with genetic screening can already prevent transmission of many heritable conditions. Bioethicist Hank Greely warned that affluent individuals could set up a private IVF and genetic-testing lab “for probably a handful of millions of dollars” to pursue embryo editing. Others judged the work more careful than prior efforts: Greg Neely told Nature it was “less reckless, more careful and ethical than previous attempts.” This is a research report, not clinical guidance.

TRANSCEND-T2D-1 was a 40-week, double-blind, randomised, placebo-controlled trial run at 48 sites in the USA, Mexico, and India, enrolling adults whose diabetes was inadequately controlled by diet and exercise alone (Lancet, DOI 10.1016/S0140-6736(26)00967-0). Of 537 participants randomised 1:1:1:1, 134 received retatrutide 4 mg, 133 received 9 mg, 136 received 12 mg, and 134 received placebo, all by once-weekly injection. Baseline mean HbA1c was 7.9% and mean BMI was 35.8 kg/m².

Who was studied — a narrow, drug-naive group

This was not a typical clinic population. Per the ClinicalTrials.gov registry, eligibility was restricted to adults with an HbA1c of 7.0–9.5% who were insulin-naive and had taken no oral or injectable antihyperglycaemic medication for at least 90 days before screening, with a BMI of 23.0 kg/m² or higher (NCT06354660). In other words, these were people with relatively early, untreated type 2 diabetes — mean diabetes duration was just 2.5 years — managing on diet and exercise alone. The results say nothing about people already on metformin, other agents, or insulin, or about longer-standing or more-advanced disease.

What the trial showed

On the primary endpoint — change in HbA1c at week 40 — retatrutide cut HbA1c by 1.69%, 1.86%, and 1.94% across the three doses, versus 0.81% with placebo. That translated to placebo-adjusted reductions of 0.88% (95% CI −1.18 to −0.59), 1.04% (−1.32 to −0.76), and 1.12% (−1.39 to −0.85), all p<0.0001.

Weight loss was a SECONDARY endpoint: bodyweight fell 11.5%, 13.9%, and 15.3% across doses, against 2.6% with placebo — a notable signal, but not the trial’s primary outcome and not an approved use.

Safety, exclusions, and known limitations

The most frequent adverse events were generally mild-to-moderate gastrointestinal events that subsided over time. Adverse events led 2–5% of retatrutide-treated patients to discontinue, versus 0% of placebo-treated patients; no participant had severe hypoglycaemia. Two patients died, both in the 4 mg group, and investigators judged both deaths unrelated to study drug.

The trial deliberately excluded people who carry the highest safety risks for this drug class, so the favourable safety picture should be read with those exclusions in mind. Per the registry, the study barred anyone with a personal or family history of medullary thyroid carcinoma (MTC) or multiple endocrine neoplasia syndrome type 2 (MEN2), a history of acute or chronic pancreatitis, severe hypoglycaemia within the prior 6 months, NYHA class IV heart failure, an acute myocardial infarction, stroke, or heart-failure hospitalisation within 90 days, or any lifetime history of a suicide attempt. The MTC/MEN2 exclusion mirrors the boxed-warning-class caution carried by other incretin (GLP-1) agonists, which are contraindicated in people with that history; this remains a known limitation and contraindication consideration for the drug class.

Per the registry, the Eli Lilly–sponsored trial is completed (NCT06354660). Because this was monotherapy in early, drug-naive disease, generalisability to longer-standing or treated diabetes remains to be established.

This is a research result, not medical guidance. Retatrutide is not available by prescription for type 2 diabetes or any other indication, and nothing here should be read as a reason to start, switch to, or ask a clinician to prescribe it. Treatment decisions belong with a patient and their own physician.

FIND-CKD (NCT05047263), a Bayer-funded phase 3 trial, randomized 1,584 adults with non-diabetic CKD, albuminuria, and a renin-angiotensin system inhibitor on board: 793 to finerenone and 791 to placebo, per the New England Journal of Medicine report. The trial enrolled a population deliberately selected for lower hyperkalemia risk: per the registry, participants had to have a baseline serum potassium of 4.8 mmol/L or lower, be on a stable, maximally tolerated labelled dose of an ACE inhibitor or ARB for at least four weeks, and not have diabetes (type 1 or 2 diabetes, or an HbA1c of 6.5% or higher, was an exclusion). Those entry criteria matter for reading the results — they cannot simply be generalized to unselected patients.

The primary endpoint — the total eGFR slope, the mean annual change in estimated glomerular filtration rate from baseline to month 32 — favored the drug. eGFR fell by 3.3 mL/min/1.73m² per year on finerenone versus 4.0 on placebo, a difference of 0.7 (95% CI, 0.3 to 1.1; P<0.001).

A prespecified composite of kidney-or-cardiovascular events was also lower with finerenone (hazard ratio 0.77; 95% CI, 0.60 to 0.99; P=0.043). The signal narrows on the components: HR 0.78 (0.60 to 1.01) for the two kidney events and 0.60 (0.27 to 1.33) for the two cardiovascular events — both confidence intervals crossing 1. Hyperkalemia was the most common adverse event: 17.0% of finerenone-treated participants versus 13.3% of placebo-treated participants had it, and it led to hospitalization in 0.9% and 0.6%, respectively — a rate to read against those potassium and background-therapy entry rules, not as a population-wide estimate. Finerenone is contraindicated or hazardous in the setting of hyperkalemia and with certain concomitant drugs (such as strong CYP3A4 inhibitors), and serum potassium and kidney function require monitoring; this is not a treatment decision readers should self-direct.

What the pool adds

The INFINITY analysis, published in The Lancet, combines individual participant data from FIDELIO-DKD, FIGARO-DKD, and FIND-CKD — 14,574 participants across diabetic and non-diabetic CKD. Finerenone cut the composite kidney outcome (kidney failure or a sustained eGFR drop of 57% or more) by 24% versus placebo (HR 0.76; 95% CI, 0.68 to 0.86) and kidney failure alone (HR 0.85; 0.74 to 0.99). The cardiovascular composite fell to HR 0.80 (0.70 to 0.91), with all-cause death at 0.88 (0.79 to 0.99).

Treatment effects on the kidney outcome were consistent regardless of glycaemic status, CKD aetiology, baseline eGFR, albuminuria, or SGLT2-inhibitor use, the authors report.

The trials enrolled selected, albuminuric populations on background therapy with screening potassium ceilings; hyperkalemia remains the trade-off. The authors frame finerenone as a “foundational therapy” across CKD — but for non-diabetic CKD specifically, that is an investigational claim, not a current label. Finerenone is not yet approved for CKD without diabetes; any expansion of its use would await regulatory review, and it is regulators, not the journal article, who will adjudicate both the approval status and the “foundational therapy” framing.

A note on who was studied. Per the trial registry, ACCESS excluded anyone with a previous documented diagnosis of diabetes mellitus and enrolled overweight participants (BMI ≥27) only if they had at least one weight-related comorbidity. The results therefore should not be assumed to generalize to people with diabetes, or to overweight people without a weight-related comorbidity.

ACCESS randomized 230 adults with obesity or overweight — mean body-mass index 39.5 kg/m², 54% female — to once-daily aleniglipron, escalated every four weeks to 45, 90 or 120 mg, or to placebo, in a double-blind design. Per ClinicalTrials.gov, the study (NCT06693843), sponsored by Gasherbrum Bio, a subsidiary of Structure Therapeutics, ran across 39 US sites and is now active but not recruiting, with an open-label extension underway.

The primary endpoint was percent change in body weight from baseline to week 36. At that point, the placebo-adjusted least-squares mean body-weight change was −8.2% (95% CI −11.1 to −5.3) for 45 mg, −9.8% (−12.5 to −7.2) for 90 mg and −11.3% (−13.9 to −8.6) for 120 mg, all significant at P < 0.0001 versus placebo. The authors reported no apparent weight-loss plateau by the end of the 36-week double-blind period — an interim observation from a dose-finding trial, not a durability finding — and continued weight loss at an interim analysis of the open-label extension (median treatment duration 20 weeks).

“Clinically relevant weight reductions of up to 11.3% with a tolerability profile consistent with the GLP-1RA class support further development.”

How it compares

On safety, the authors reported that gastrointestinal events among aleniglipron-treated participants were generally mild to moderate and decreased over time, with little to no recurrence of vomiting after dose reintroduction following permitted interruptions. Treatment-related discontinuations reached 10.4% across the aleniglipron arms, and the authors recorded no events of drug-induced liver injury in aleniglipron-treated participants — a point of interest given liver-signal concerns that have dogged some oral small-molecule candidates in this class.

Two caveats temper the readout. This is a 36-week, dose-finding phase 2b trial, not a confirmatory phase 3, and the headline figures are placebo-adjusted, not absolute. The deeper responder-rate breakdowns and head-to-head positioning against injectable semaglutide or tirzepatide, and against oral rivals such as orforglipron, await larger and longer trials. For now, ACCESS establishes that a small-molecule pill can reach double-digit weight loss over 36 weeks without an early plateau in this dose-finding population.

In the new trial, also in Nature Medicine, seven participants with profound, chronic upper-limb deficits — baseline Fugl-Meyer Assessment scores of 15–35 out of 66 — were implanted with two leads placed unilaterally in the cervical spinal cord for four weeks, then explanted. (The registered eligibility window was broader, Fugl-Meyer above 7 and below 50; 15–35 describes the cohort that actually enrolled, not the cutoff.) Per the registry (NCT04512690), the primary endpoints were safety and tolerability — serious adverse events, and stimulation-related discomfort or pain — not efficacy. On that score it cleared: no serious adverse events occurred. Fugl-Meyer was a secondary outcome.

The motor signals were the headline. With stimulation switched on, function improved immediately regardless of severity — an average +32% in strength and +5.6 Fugl-Meyer points versus each participant’s own baseline. Three of the seven, all with residual corticospinal connectivity to finger muscles, regained some hand and finger movement under stimulation.

Despite just 8.6 hours of motor activity over the month (5.5 with stimulation on), participants gained an average of +6.6 Fugl-Meyer points at the end of the study versus baseline — and spasticity eased in all seven. The published report gives point estimates only; no confidence intervals, standard deviations or p-values are reported for these aggregate effects.

What this is, and isn’t

This is a seven-person, single-arm feasibility study with a temporary implant — the authors call it “preliminary evidence” of safety, feasibility and efficacy. There is no control group, the gains were measured during stimulation, the implant came out within 30 days, and the headline figures are reported without dispersion or significance testing. The end-of-study Fugl-Meyer change of +6.6 points sits within the 4.25–7.25-point range usually considered clinically meaningful (the MCID) and above the 5.2-point minimal detectable change the registry cites — but with n=7, no comparator and no reported confidence interval, the estimate carries wide uncertainty. What the trial establishes is feasibility and a safety signal — the foundation, the authors note, for larger efficacy studies of a potentially fully implantable neuroprosthetic.

The trial, published in The New England Journal of Medicine on June 2 and presented at EULAR 2026, randomized 194 adults with active IgG4-related disease 1:1 to subcutaneous obexelimab 250 mg or placebo once weekly for 52 weeks, with glucocorticoids tapered to discontinuation by week 8 in both groups. The primary endpoint was time to the first flare requiring rescue therapy, adjudicated by both the investigator and an independent committee.

On that endpoint, obexelimab reduced the risk of flare by 56% (hazard ratio 0.44; 95% CI, 0.28 to 0.71; P<0.001). Flares requiring rescue occurred in 26 of 97 patients (26.8%) on obexelimab versus 53 of 97 (54.6%) on placebo.

Weekly obexelimab led to a significantly lower risk of disease flare and significantly less glucocorticoid exposure than placebo.

The drug also met its key secondary endpoints. Complete remission at week 52 reached 37.1% with obexelimab versus 19.6% with placebo (P=0.005), and the cumulative dose of rescue glucocorticoid was 329.5 mg versus 929.8 mg (P=0.004) — roughly a two-thirds reduction in steroid exposure, the practical goal in a steroid-sparing disease.

Mechanism and safety

Unlike rituximab and other B-cell-depleting agents used off-label here, obexelimab is a bifunctional antibody that co-engages CD19 and the inhibitory receptor FcγRIIb to dampen B-cell activity without depleting the cells. Across the safety data, more obexelimab-treated than placebo-treated patients reported arthralgias (19.6% vs. 11.3%), hypersensitivity (16.5% vs. 11.3%), and diarrhea (11.3% vs. 6.2%); serious adverse events were less frequent on obexelimab, affecting 10.3% of obexelimab-treated patients versus 18.6% of placebo-treated patients.

Obexelimab remains investigational and is not approved for any use. Sponsor Zenas BioPharma says it submitted a Biologics License Application to the FDA in May; that claim comes from the company and has not been confirmed by the agency. A BLA submission begins the FDA’s review process — it is not an approval and does not make the drug available as standard care. No therapy is yet approved specifically for IgG4-related disease. The trial’s peer-reviewed protocol paper describes INDIGO, which enrolled 194 patients, as the largest Phase 3 trial in IgG4-related disease to date; it continues into a three-year open-label extension.

The trial randomized 381 patients with recently relapsed disease in a 1:1:1 ratio to secukinumab 300 mg, secukinumab 150 mg, or placebo for 52 weeks, with 127 patients in each group. Every patient also received prednisone on a 24-week tapering schedule, so the comparator was an active steroid taper, not nothing.

The remission gap

The primary endpoint was sustained remission at week 52: remission achieved by week 12 and maintained through week 52 without escape or rescue treatment and with no new diagnosis of giant-cell arteritis. Sustained remission was reached by 41.2% of patients on 300 mg (95% CI, 32.8 to 49.7) and 40.6% on 150 mg (95% CI, 32.2 to 49.0), versus 20.4% on placebo (95% CI, 13.6 to 27.2). Each dose beat placebo at P<0.001. Put the other way, even on the active doses roughly 6 in 10 patients did not reach sustained remission.

Both doses roughly doubled the one-year sustained-remission rate over a steroid taper alone — and the two doses landed within a percentage point of each other.

The steroid-sparing signal followed, but as a secondary endpoint. The mean adjusted annual cumulative glucocorticoid dose was 1603.7 mg on 300 mg and 1683.2 mg on 150 mg, against 2093.0 mg on placebo — roughly a 400-to-490 mg reduction over the year, on this defined 24-week taper and in this relapsed population.

Safety looked broadly balanced. Serious adverse events occurred in 13.5% of patients on 300 mg, 15.9% on 150 mg, and 14.2% on placebo. Consistent with the IL-17 class, more secukinumab-treated than placebo-treated patients had nasopharyngitis, hypersensitivity reactions, urinary tract infections, fungal infections, and back pain.

Two caveats temper the enthusiasm. The trial enrolled patients who had already relapsed while tapering glucocorticoids, so the findings speak to harder-to-control disease rather than first-line use. And the near-identical performance of the two doses, while convenient, leaves the dose-response question open. This is a regulatory-grade efficacy signal for the first IL-17A-targeted therapy in PMR — not, on its own, a treatment recommendation.

The investigator-initiated, quadruple-blind trial (NCT03969953) randomized 1,458 adults with CKD stage 4-5 (eGFR 29 mL/min/1.73 m2 or lower) or dialysis-dependent kidney failure, roughly half of them on dialysis, to rivaroxaban (Xarelto) 2.5 mg twice daily or placebo. All carried elevated cardiovascular risk. Over a median 1.7 years, the primary composite of cardiovascular death, nonfatal MI, nonfatal stroke, or nonfatal peripheral artery disease event occurred in 164 rivaroxaban patients (22.6%) versus 151 on placebo (20.7%) — a hazard ratio of 1.09 (95% CI, 0.87-1.36; P=.46).

There was no signal of benefit on any major secondary endpoint, including all-cause mortality (HR 1.14; 95% CI, 0.92-1.40). The trial was stopped early after the data monitoring board, at a prespecified interim review, recommended termination on July 8, 2025, citing “concerns regarding net harm and a low probability of demonstrating efficacy”; the steering committee accepted that recommendation on August 7, 2025. The article is explicit that the decision was not based on prespecified stopping rules. It rested instead on a post hoc analysis estimating a conditional power of just 16% for an assumed hazard ratio of 0.78.

More bleeding, no payoff

The harm was clearer than any benefit. Major bleeding, the primary safety outcome, occurred in 64 rivaroxaban-treated patients (8.8%) versus 44 placebo-treated patients (6.0%) — HR 1.51 (95% CI, 1.02-2.22; P=.04).

“Clinical decisions regarding anticoagulation in people with advanced kidney disease should not be extrapolated from other cardiovascular populations,” the authors wrote.

One mechanistic clue: 42.1% of deaths were sudden cardiac deaths, which the authors note may not be atherothrombotic — and so may sit beyond any anticoagulant’s reach. A nominal reduction in venous thromboembolism (HR 0.29; 95% CI, 0.09-0.87) sat outside the trial’s multiplicity-controlled testing hierarchy and carries no valid P value (reported as “NA”). After a missed primary endpoint, it is an exploratory, hypothesis-generating observation, not an established effect, and does not rescue the primary result.

ANVIL (NCT02595944), sponsored by the National Cancer Institute, was a randomized phase 3 study run at 378 sites in the US National Clinical Trials Network. Between May 2016 and September 2019 it enrolled patients with resected stage IB (at least 4 cm), II, or IIIA non-small cell lung cancer without sensitizing EGFR or ALK alterations, all of whom had completed planned adjuvant chemotherapy and/or radiotherapy. In total, 935 patients were randomized: 466 to nivolumab (480 mg intravenously every 4 weeks for up to one year) and 469 to observation.

No benefit on the co-primary endpoint

Disease-free survival, a co-primary endpoint alongside overall survival, showed no advantage. Median disease-free survival was 71.3 months with nivolumab versus 68.8 months with observation (hazard ratio for progression or death, 0.97; 97% CI, 0.79-1.20; 1-sided P = .39). In the prespecified subgroup with PD-L1 expression of 50% or higher, the hazard ratio was 0.86 (98% CI, 0.55-1.34; P = .22), also not significant.

“Adjuvant nivolumab was not associated with improved disease-free survival,” the authors concluded.

Because overall survival was to be tested only if disease-free survival was positive, it was not formally analyzed; an exploratory look showed a hazard ratio of 1.02 (95% CI, 0.82-1.26) in the overall population after a median follow-up of 72.6 months. Grade 3 to 5 treatment-related toxicities occurred in 116 nivolumab patients (25%), including two fatal respiratory events.

The result contrasts with adjuvant immunotherapy regimens that have won approval, and the authors note that nivolumab here was given after, rather than concurrent with, chemotherapy. The wide confidence interval crossing 1.0 leaves little room for a hidden benefit. Association is not causation, and these data do not constitute medical advice.

Patients were screened at 75 hospitals in China and Thailand between November 2019 and April 2024. Of 1646 screened, 588 with resectable, PD-L1-positive gastric or gastro-oesophageal junction adenocarcinoma — a combined positive score (CPS) of 5 or higher — were randomly assigned, all at 57 hospitals in China. The serplulimab group (n=292) received neoadjuvant serplulimab plus S-1 and oxaliplatin (SOX), then adjuvant serplulimab; the placebo group (n=296) received placebo plus SOX, then adjuvant SOX alone.

What the trial found

The primary endpoint was investigator-assessed event-free survival — time from randomisation to progression, local or distant recurrence, a new malignancy, or death. Following a prespecified hierarchical testing sequence, efficacy was evaluated first in the most PD-L1-rich patients (CPS ≥10) and then in the full intention-to-treat (CPS ≥5) population.

In the CPS ≥10 population, the first prespecified analysis, median event-free survival was not reached with serplulimab versus 42.0 months with placebo (hazard ratio 0.65, 95% CI 0.47–0.90; p=0.0082). In the intention-to-treat (CPS ≥5) population, tested next, median event-free survival was again not reached with serplulimab versus 35.9 months with placebo (hazard ratio 0.73, 95% CI 0.56–0.94; p=0.015).

Grade 3-or-worse treatment-related adverse events were less common with serplulimab — 47% versus 59% — and fewer patients discontinued treatment.

The authors note that overall survival data are not yet mature and that extended follow-up “is warranted to confirm a survival advantage.” Longer event-free survival is a meaningful but intermediate signal; whether it translates into patients living longer remains unproven. And although patients were screened in China and Thailand, every randomised patient was enrolled in China, so applicability to other populations is untested. The study was funded by Shanghai Henlius Biotech, which makes serplulimab.