FDA spells out its psychedelic framework

And a re-read of 1965-77 trials argues durable benefit was never really there.

Weekly Science Digest, 5 Sep 2026 to 11 Sep 2026. 37 papers read, 6 preprints, 1 new trials.

1. Regulators write for a general medical audience

Four FDA staff, including the psychiatry division director, set out the agency's psychedelic framework in NEJM.

It lands the same week JAMA ran both an Insights piece on psilocybin for depression and a patient-facing explainer, which is its own signal about where the audience now is.

2. The 1960s data, re-read

A review of 11 controlled trials from 1965-77 concludes psychedelic therapy produced no durable benefit past a few weeks.

Passie and van Elk also push back on the received story that those studies were methodologically hopeless, arguing they handled blinding and active controls about as well as contemporary trials do.

3. This week's human data

MDMA meta-analysis stays positive; a psilocybin smoking cohort cut craving but not abstinence.

Pooling seven RCTs (published and unpublished, none retracted), MDMA-assisted therapy improved PTSD severity, dissociation, depression and functional impairment, but not sleep. In an observational smoking cohort, psilocybin-assisted care beat standard care on reduction and early urge scores while the adjusted abstinence odds ratio was 1.82 (95% CI 0.68-4.86).

4. Taking the trip out, and an argument against

A PNAS study strips LSD down to its minimal pharmacophore and finds safer, less hallucinogenic analogues.

Nine simplified ergolines were profiled across 5-HT2A/2B/2C; UCD0076 prefers 5-HT2C, shows antipsychotic-like effects in mice and avoids LSD's cardiotoxic liability. A separate report has a non-hallucinogenic LSD analogue boosting anti-tumour immunity in colorectal cancer models.

5. In brief

6. In the clinic

How much weight it carries: Trial registry movements: what is being attempted, not what was found. (Registry, Plans, not results)

7. On the preprint servers

How much weight it carries: Not peer reviewed. Early results that may not hold up. (Preprint)

—
Node—
Statuslive
Throughput—