The drug
COMP360 is synthetic crystalline psilocybin, given as a single 25 mg oral capsule. Orange Book patents cover the drug substance, the drug product and the method of treatment, with exclusivity expected beyond 2038.
- How it is given: at least six hours of monitoring by a licensed provider at a certified centre, with psychological support.
- Indication: treatment-resistant depression (TRD), defined as failure of two to four adequate treatments in the current episode (MGH-ATRQ).
- Designations: Breakthrough Therapy (October 2018) and the UK’s ILAP pathway.
- Market: about 4 million US adults have TRD (Compass, citing an ADAA 2026 poster). Spravato is the only drug approved and used for it.
- Burden of disease: episodes last about 571 days, twice as long as non-resistant depression, with 70% more lost work time, 51% higher suicide mortality and 17–30% higher all-cause mortality.
Pipeline beyond TRD. A Phase 2b/3 trial in PTSD (COMP202, n=300, 25 mg ×2 vs. 10 mg vs. 1 mg, primary CAPS-5 at week 8) started in September 2026, with primary data around November 2028; an earlier open-label Phase 2 (n=22) was positive. Phase 2 studies are also running in recurrent MDD (n=102) and completed in anorexia (n=32). In practice Compass is a single-asset, single-indication story for the next two years.
Mechanism
Psilocybin is a prodrug: the body converts it to psilocin, a direct agonist of the serotonin 5-HT2A receptor (ChEMBL CHEMBL194378). The FDA’s July 2026 guidance defines classic psychedelics as 5-HT2A agonists and describes rapid, durable benefit as a hypothesis rather than an established fact.
- Type of drug: symptomatic and procedural. It has not been shown to be disease-modifying.
- Interactions: the FDA notes that SSRIs, SNRIs or MAOIs taken for three weeks or more may blunt the effect. Both trials required patients to stop prohibited medications, which matters for how the drug will be used in practice.
- Class safety flag: the FDA flags 5-HT2B agonism, a heart-valve risk with chronic dosing. Compass has not disclosed its 5-HT2B data.
The two Phase 3 trials
| COMP005 (NCT05624268) | COMP006 (NCT05711940) | |
|---|---|---|
| Design | Randomised, double-blind, placebo-controlled | Randomised, double-blind, dose-controlled |
| Patients dosed | 258: 25 mg 171 / placebo 87 (2:1) | 581: 25 mg 296 / 10 mg 142 / 1 mg 143 (2:1:1) |
| Dosing | 1 dose | 2 doses, 3 weeks apart |
| Primary endpoint | MADRS at week 6, 25 mg vs. placebo | MADRS at week 6, 25 mg vs. 1 mg |
| Registered secondaries | Sheehan Disability Scale (SDS), week 6 | SDS; 25 vs. 10 mg on MADRS and SDS, week 6 |
| Key secondary (Compass) | — | MADRS at week 9 |
| Blinded Part B | Weeks 6–26, same-arm re-dose allowed | Weeks 9–26, same-arm re-dose allowed |
| Part C, open label | Weeks 26–52, completed | Weeks 26–52, completion Dec 2026 |
| Max doses in 52 weeks | 3 | 4 |
| Primary data | 28 May 2025 | 13 November 2025 |
| Sites | US only, 32 dosing sites | North America and Europe, 116 listed |
Who was enrolled. Adults with major depression without psychotic features, MADRS ≥20 at screening and baseline. Exclusions were broad: bipolar or psychotic disorders, several personality disorders, ongoing PTSD, OCD or anorexia, a psychiatric hospital stay in the past 12 months, ECT, DBS or VNS in the current episode, TMS within six months and any prior COMP360.
Baseline. These were seriously ill, chronic patients. Mean MADRS was 31.5–32.5 across arms; the current episode had lasted 35–46 months on average; patients averaged five to eight lifetime episodes; and 64–74% had failed two treatments in the current episode. Prior psychedelic use was low at 4.7–6.9% per arm (the cap was 15%), which matters for blinding.
Efficacy: every timepoint
Gap versus control in MADRS points, with the 95% confidence interval and p-value. Negative is better.
| Timepoint | COMP005: 25 mg vs. placebo | COMP006: 25 mg vs. 1 mg | COMP006: 10 mg vs. 1 mg |
|---|---|---|---|
| Day 2 | −4.7 (−7.0, −2.3) p<0.001 | −2.9 (−4.6, −1.2) p<0.001 | −3.0 (−5.0, −1.0) p=0.002 |
| Week 1 | −7.2 (−9.6, −4.8) p<0.001 | −5.0 (−7.0, −3.0) p<0.001 | −3.5 (−5.9, −1.2) p=0.001 |
| Week 3 | −5.2 (−7.4, −2.9) p<0.001 | −4.4 (−6.3, −2.5) p<0.001 | −3.4 (−5.6, −1.2) p=0.001 |
| Day 24 | — | −5.0 (−7.1, −2.9) p<0.001 | −4.9 (−7.4, −2.5) p<0.001 |
| Week 4 | — | −4.8 (−6.9, −2.7) p<0.001 | −4.9 (−7.3, −2.5) p<0.001 |
| Week 6 (primary) | −3.6 (−5.7, −1.5) p<0.001 | −3.8 (−5.8, −1.8) p<0.001 | −2.5 (−4.8, −0.2) p=0.017 |
| Week 9 | No statistics after week 6 | −2.9 (−5.0, −0.9) p=0.002 | −2.6 (−5.0, −0.2) p=0.016 |
Values from the August 2026 investor deck, the latest document with full p-values. The February deck gave slightly different figures for some timepoints; we use the later ones.
What the table says. Every timepoint is significant in both trials, and the 10 mg dose separates too, which gives a clean dose response. But the shape matters: the effect peaks at week 1 and then narrows. In COMP005 the gap falls from −7.2 at week 1 to −3.6 at week 6. In COMP006 it falls from −5.0 to −3.8 at week 6 and −2.9 at week 9.
In COMP006, arm-level improvement from baseline at week 6 was −7.6 on 25 mg, −6.3 on 10 mg and −3.7 on 1 mg; by week 9 it was −5.6, −5.3 and −2.7.
Response and remission
Compass reports responders on a soft, post hoc definition (a drop of at least 25% in MADRS), rather than the standard 50%.
- COMP005, 25 mg arm: 42 of 171 patients (25%) improved by at least 25%; 115 did not. Remission was 6%.
- COMP006, 25 mg arm: 116 of 296 (39%) improved by at least 25%; 153 did not. Remission was 16%.
- Among responders who had not remitted, more than 40% (COMP005) and 28% (COMP006) remitted after an extra dose.
- The standard ≥50% response rate has not been disclosed for the controlled parts of either trial.
Not disclosed: the Sheehan Disability Scale (a registered secondary) in both trials, the 25 vs. 10 mg comparison, any controlled 12-week result, and control-arm response and remission rates.
Re-dosing and the open-label year
After week 6, 70% of 25 mg patients in COMP005 were re-dosed against 53% of placebo patients; in COMP006, 58% of 25 mg patients were re-dosed after week 9. Antidepressants were allowed again in Parts B and C.
About 70% of patients entered the open-label Part C. In the 25 mg arm, around 80% took another dose and improved by a mean 13 points from baseline at week 52; around 90% of the placebo crossover group took a first 25 mg dose and improved by 10 points. After the Part C dose, 40–45% responded on the standard ≥50% definition and around 30% remitted. These are open-label results and should be read as supportive, not confirmatory.
Safety
Most side effects are what you would expect from a psychedelic session: they start on the dosing day (89.5%) and resolve within a day (86.1%).
| COMP006 through week 26 | 25 mg | 1 mg |
|---|---|---|
| Nausea | 44.9% | 12.6% |
| Headache | 39.9% | 34.3% |
| Anxiety | 28.0% | 20.3% |
| Visual hallucination | 17.2% | 3.5% |
| Illusion / perceptual disturbance | 16.2% | 2.8% |
| Fatigue | 16.6% | 13.3% |
| Dizziness | 15.9% | 6.3% |
| Crying | 13.9% | 4.2% |
| Blood pressure increased | 11.8% | 2.8% |
Serious adverse events through week 26
| Event | 005 · 25 mg | 005 · placebo | 006 · 25 mg | 006 · 10 mg | 006 · 1 mg |
|---|---|---|---|---|---|
| Any serious AE | 8 (4.7%) | 2 (2.3%) | 17 (5.7%) | 5 (3.5%) | 9 (6.3%) |
| Suicidal ideation | 4 (2.3%) | 0 | 4 (1.4%) | 2 (1.4%) | 4 (2.8%) |
| Depression, suicidal | 1 | 0 | 0 | 0 | 0 |
| Suicidal behaviour | 0 | 0 | 0 | 0 | 1 |
| Suicide attempt | 0 | 0 | 0 | 0 | 1 |
| Suspected suicide | 0 | 0 | 1 (judged unrelated) | 0 | 0 |
| Anaphylactic reaction | 1 | 0 | 0 | 0 | 0 |
Pooled across both trials, serious suicidal ideation was identical: 8 of 467 (1.7%) on 25 mg against 4 of 230 (1.7%) on placebo or 1 mg. The imbalance in COMP005 alone is a labelling and REMS issue rather than an approval risk.
Competitors
| COMP360 (Compass) | Spravato (J&J) | DT120 / LSD (Definium) | |
|---|---|---|---|
| Status | NDA rolling, decision ~mid-Jan 2027 | Approved; $1.5B 2025 revenue | Phase 3 (Emerge read out Jun 2026) |
| Effect vs. control | −3.6 / −3.8 at week 6 | −4.0 (with antidepressant, week 4); −5.1 / −6.8 monotherapy, day 28 | −8.1 at week 6; −7.3 at week 12 |
| Control | Placebo; 1 mg active dose | Placebo | Inert placebo only |
| Remission | 6% / 16% (week 6) | — | 24% vs. 3% |
| Durability | 76% of effect kept week 6 → 9 | Relapse HR 0.30–0.49 in maintenance | 90% kept week 6 → 12 |
| Sessions in year one | 1–4 (6–24 hours) | 34–56 (68–112 hours) | — |
Spravato is the incumbent: fewer than 100,000 patients, more than 8,100 certified clinics, twice-weekly dosing in the first month, then weekly, then every one to two weeks, with at least two hours of monitoring each time. Its effect is similar to COMP360’s, but the treatment burden is an order of magnitude higher.
DT120, Definium’s LSD tablet, posted a much larger effect (−8.1 at week 6, response 35% vs. 7%, remission 24% vs. 3%, no serious adverse events) in a 149-patient trial. Two caveats: the control was an inert placebo, which is easier to beat when patients can tell what they took, and the second Phase 3 (Ascend) does not read out until 2027. It is the competitor that matters most on efficacy, but it is at least a year behind.
HLP003 (Helus, formerly Cybin’s CYB003) has a 223-patient trial with topline data due in Q4 2026. We have only secondary sources on it so far.
Usona filed its own psilocybin for approval on 2 November, also with a priority voucher. This is the competitor that moved the stock, because it threatens Compass’s first-to-market position rather than its clinical profile.
Burden, not speed, is the edge
- Onset: no advantage. COMP360, Spravato and DT120 all separate from control by day 2. Compass’s own comparison slide used only Spravato’s pre-approval trials, which made “faster onset” look true when it is not.
- Burden: in year one, COMP360 needs 8–56× fewer visits and 3–19× fewer hours than Spravato; for a typical patient, 11–17× fewer visits and 4–6× fewer hours. That matters to patients, to clinics with limited treatment rooms and potentially to payers.
- The catch: the advantage only applies to responders, and 61–75% of 25 mg patients did not reach even a 25% improvement by week 6.
How COMP360 scores against the FDA’s psychedelic guidance
The FDA finalised its guidance on psychedelic drug development in July 2026. Scoring Compass against it is the best way to judge approval risk.
| FDA recommendation | Status |
|---|---|
| Low-dose control | Met 1 mg and 10 mg arms |
| Dose response | Met |
| Prior psychedelic users not over-represented | Met 5–7% |
| 12-week double-blind effect before a first MDD application | Open not disclosed |
| Blinded long-term follow-up with pre-set re-dosing criteria (~12 months) | Partly met 26 weeks |
| Central blinded raters; blinding and expectancy questionnaires | Unknown |
| Results “strongly persuasive and robust across endpoints” | Open SDS not disclosed |
| Isolate the contribution of psychotherapy | Not done |
| Two monitors per session; REMS | In progress |
Regulatory path and timeline
- April 2026: rolling NDA submission granted, plus a National Priority Voucher; Compass cites a possible one-to-two-month review.
- 3 November 2026: management said the filing would be complete within two weeks, citing a median review of about 60 days.
- ~Mid-January 2027: our expected FDA decision. There is no formal PDUFA date.
- First half of 2027: launch guided, subject to approval and DEA rescheduling. States covering about 90% of US patients plan to reschedule within 30 days of the DEA, and a White House executive order directs the DEA to move quickly.
- Funding: $433 million of cash against $51 million of debt funds the company into 2028, past launch.
Where we are sceptical
- The effect peaks early and fades. The gap is largest at week 1 and narrows to week 9. Strong concern
- Functional outcomes are missing. The Sheehan Disability Scale was registered and has not been disclosed. Moderate
- Soft responder definition and low remission. A 25% threshold flatters the numbers; remission is 6% and 16%. Strong concern
- Durability is post hoc. The long-term data are open-label or rest on re-dosing. Strong concern
- Functional unblinding. Patients can usually tell whether they had a psychedelic experience. The dose response mitigates this, but it remains the main approval risk.
- COMP005 suicidality imbalance. Balanced when pooled; a label and REMS issue.
What holds up: two trials, 839 patients, every timepoint significant and a dose response. That is a stronger statistical package than most psychiatry approvals.
Why we own the calls
The market has priced Usona’s filing as if Compass has lost its head start. We think that is too pessimistic. Compass filed first, has the larger and replicated Phase 3 package, and its burden advantage over Spravato is real. At about $9, below the pre-Usona level of $11.70, the shares do not reflect an approval that we think is likely. Our target is $14–16 after approval, still below the median analyst target of about $21.
Because an FDA decision is close to binary, we express the view through $10 calls rather than stock: the loss is capped at the premium and sizing is deliberately small. Full trade details are in the 10 November letter.
Sources
- Compass Pathways August 2026 investor deck and 17 February 2026 Phase 3 presentation
- COMP005 52-week, COMP006 26-week and COMP006 primary press releases (ir.compasspathways.com)
- Compass Q2 2026 10-Q (SEC EDGAR)
- FDA, Psychedelic Drugs: Considerations for Clinical Investigations, final guidance, July 2026
- Spravato prescribing information, 2025 (FDA label 211243s016)
- Definium Therapeutics, Emerge Phase 3 press release, 22 June 2026
- ChEMBL CHEMBL194378 (psilocybin), CHEMBL65547 (psilocin); ClinicalTrials.gov NCT05624268, NCT05711940, NCT07570654