Approval of an mRNA flu vaccine, survival in an intracranial CAR-T trial, and a symptom-score gap in a psilocybin study mark different evidence thresholds for interventions.
Science··Evening
Approval carries two meanings across age groups
Moderna's mFLUSIVA became the first influenza vaccine to use mRNA technology through an FDA decision dated 5 August. Its scope is divided: adults aged 50-64 received full approval, while people aged 65 and older entered through the accelerated pathway. Scientific American's report therefore places two evidence states under one product name. Full approval says the application for the 50-64 group crossed the ordinary review threshold. For the older group, accelerated approval rests on a surrogate endpoint considered reasonably likely to predict clinical benefit; patient outcomes have not yet been confirmed. The FDA required Moderna to conduct an additional confirmatory trial in that population. Earlier in the year, the agency had declined to review the application before reversing its decision within a week. Approval does not close every question to the same degree. A regulatory route is open, but for people aged 65 and older the surrogate measure supporting approval and the outcomes awaiting confirmation remain separate steps. That distinction keeps a regulatory milestone from being mistaken for direct demonstration of benefit in every covered population.[1]
Survival accompanies an early-phase safety study
The TX103 study in Nature Medicine reports another milestone. Autologous CAR-T cells targeting B7-H3 were delivered directly into the skull for recurrent glioblastoma. The open-label phase 1 trial used a 3+3 dose-escalation design: 15 patients received 72 infusions, and 13 received repeated infusions. Investigators found no dose-limiting toxicity or maximum tolerated dose and recommended the middle level for phase 2. In this group, 12-month overall survival was 66.7 per cent and median overall survival was 19.1 months after the first infusion. The findings are an important early signal, but the phase and design bound their meaning. The study did not test a survival advantage against a control arm; it primarily opened a dose and safety range. The 95 per cent confidence interval for median survival begins at 8.93 months without reaching an upper bound, showing the uncertainty around one summary figure in a sample of 15. Low-grade cytokine release syndrome was common, and three grade 3 events counted as serious. The 19.1-month median is an outcome observed in an early-phase group where repeated intracranial delivery proved feasible, providing context for phase 2 rather than controlled comparative evidence or an approval decision.[2]
A controlled symptom gap sits beside feasibility outcomes
The PsiDeR trial measures a symptom difference between randomised arms. Nature Medicine reports that it assigned 60 people with treatment-resistant major depression in a 1:1 ratio: 30 received one 25-milligram dose of psilocybin, 30 received inactive placebo, and both groups received psychological support. The adjusted Montgomery-Åsberg difference was 10.41 points at week 3 and 12.92 points at week 6. The primary outcomes, however, were feasibility measures covering recruitment, retention, and estimation of scale variance. This gives a more direct comparison than TX103's single-arm survival observation while preserving a bounded question. Of 75 people screened, 60 were randomised and 59 completed the scale at every visit. Blinding failed: everyone receiving psilocybin and 70 per cent of the placebo group correctly guessed their allocation. The authors say expectancy effects probably contributed substantially. One NHS site enrolled a highly educated sample with limited occupational diversity. The 12.92-point result is a controlled symptom-scale difference, not a licence, established long-term benefit, or a definitive result for a broader population. Together, the reports require readers to name each intervention's stage, endpoint, and comparator instead of treating every instance of progress as interchangeable.[3], [2], [1]
