One Dose of LSD for Generalised Anxiety: What the JAMA Phase 2b Actually Showed
- Phase 2b multicentre RCT, n=198 adults with moderate-to-severe GAD (HAM-A ≥20), 22 US sites, randomised to placebo or 25, 50, 100, or 200 µg of MM120 (lysergide D-tartrate) as monotherapy — no adjunctive psychotherapy.
- At week 4, only the 100 µg and 200 µg arms separated from placebo on HAM-A: LS-mean difference −5.0 points (95% CI −9.6 to −0.4) and −6.0 points (95% CI −9.8 to −2.0). The 25 µg and 50 µg arms did not.
- Visual perceptual changes (illusions, pseudo-hallucinations, visual hallucinations) occurred in 92.5% of the 100 µg group, 100% of the 200 µg group, and 10.3% of placebo. Nausea on dose day: 40% (100 µg), 60% (200 µg).
- Effect onset by day 2, sustained through week 12 in earlier reports — but the formal primary endpoint is week 4 separation, and the company-disclosed 65% response / 48% remission figures are secondary outcomes from press materials, not the JAMA primary analysis.
A Tier-1 RCT testing a single dose of LSD as monotherapy for GAD is, by itself, a marker of how far the regulatory window has shifted in three years. The result is real but narrower than the press cycle suggests. Two doses cleared placebo on the primary endpoint. Two did not. And the trial was deliberately designed without psychotherapy — which puts it in different conceptual territory from psilocybin-assisted therapy and complicates direct comparison with the depression literature.
What the data shows
The primary analysis is a dose-response model, not a head-to-head test of "LSD vs placebo". The MCP-Mod method confirmed a dose-response relationship and identified 100 µg and 200 µg as the active doses. Magnitude at week 4 is moderate: a 5-6 point HAM-A reduction over placebo, against a baseline of moderate-to-severe anxiety (HAM-A ≥20). For context, the minimal clinically important difference cited in the protocol is 2.5 points — so the effect is real, sustained for at least four weeks after a single dose, and clearly above the noise floor.
The acute experience is not subtle. At 100 µg, nine in ten participants reported visual perceptual changes; six in ten reported nausea at 200 µg. These are expected pharmacodynamic effects on dose day, not adverse signals — but they are also why this protocol cannot become "outpatient pharmacotherapy" in any conventional sense. The trial used dosing-day monitoring at psychiatric research sites, blinded central raters, and a research-grade infrastructure that does not exist in standard outpatient anxiety care.
The functional unblinding problem is the elephant. With 92.5% perceptual changes on active vs 10.3% on placebo, expectancy effects almost certainly contribute to the symptom score difference. The authors acknowledge this. The Phase 3 programme (Voyage, Panorama) will face the same limitation, and it is unlikely to be solved by an active comparator at psychedelic doses.
For your practice
For the next 12-18 months — until Phase 3 readouts in 2026 — this trial does not change a single referral pathway. There is no approved psychedelic indication for GAD. Patients who ask about it are typically responding to media coverage; the honest answer is that one positive Phase 2b across two of four doses is necessary but nowhere near sufficient for clinical use, and that GAD already has multiple evidence-based options (CBT, SSRIs, SNRIs) that most patients have not exhausted.
What is worth doing now: separate the conceptual frames in your own thinking. Psilocybin-assisted therapy (PAT) for depression rests on a model where the drug enables psychotherapeutic processing — the Phase 2b PAT meta-analysis covered earlier in the digest carousel sits in that frame. MM120 in GAD is built on the opposite premise: monotherapy, no integration sessions baked into the protocol. If both pathways succeed in Phase 3, the clinical decision tree gets more complex, not simpler. Anticipate that complexity in supervision and continuing education choices now.
Two doses worked, two did not — and the 92.5% rate of visual perceptual changes is not a side effect of the treatment, it is the treatment, which is exactly why outpatient pharmacotherapy framing does not apply.
Phase 2b is hypothesis-generating for Phase 3, not regulatory. Functional unblinding is essentially complete at active doses, which inflates the symptom-rating delta on patient-reported and rater-administered measures. Sample is 83% White; generalisability across populations is unknown. Sponsor (MindMed) employees are co-authors and contributed to design, conduct, and analysis.