Designing psychedelic trials: blinding, endpoints, and comparators in FDA and EU records, and what FDA told Lykos about MDMA

Psychedelic drug development sits at an uncomfortable intersection of pharmacology and psychotherapy, and the resulting trials break several assumptions that ordinary drug trials rely on. The single most consequential problem is blinding: a 25 mg dose of psilocybin or a therapeutic dose of MDMA produces unmistakable acute subjective effects, so participants, therapists, and often raters can guess assignment. That "functional unblinding" propagates into every efficacy read-out, and it is now the defining regulatory question for the field. The registry record from ClinicalTrials.gov, EU CTIS, and the EU Clinical Trials Register shows how sponsors have tried to engineer around the problem, and FDA's complete response letter for MDMA shows how far short those efforts fell in the agency's judgment.

The blinding problem and how sponsors have tried to control it

Sponsors have converged on a small set of design tools, visible directly in the trial records.

Dose-control (low-dose active) comparators. Rather than an inert placebo, several sponsors compare a full psychoactive dose against a low dose of the same drug intended to preserve some ambiguity about assignment. COMPASS Pathways' Phase 2 treatment-resistant depression study (P-TRD, NCT03775200) randomized 1 mg, 10 mg, and 25 mg psilocybin arms, using the 1 mg dose as the low-dose control ¹¹²⁰. The same logic appears in EU CTIS: the Groningen (UMCG) trial (2023-510488-36-01) uses 1 mg psilocybin as a low-dose control against 15 mg and 25 mg doses in a 2:1 allocation ²⁵²²⁵³.

Blinded independent raters. Because unblinded site therapists cannot be trusted to score efficacy neutrally, the MDMA Phase 3 program separated the treatment dyad from the assessment. In the randomized, double-blind Phase 3 trials (NCT03537014 and NCT04077437), a pool of independent raters scored the primary endpoint while blinded to visit number and to treatment received ^51555969. The EU CTIS Groningen protocol goes further, describing all patients, caregivers, therapists, investigators, and raters as blind to the dosing condition ²⁵².

Active or masking placebos. Some programs use pharmacologically active placebos to mimic side effects and blunt guessing. In EU CTIS, a Paris (GHU Paris Psychiatrie & Neuroscience) psilocybin trial (2024-512911-34-00) pairs each arm with trazodone 30 mg, comparing psilocybin-trazodone against placebo-trazodone ^249250251. On the EU Clinical Trials Register, the Central Institute of Mental Health's Phase 2 psilocybin trial in treatment-resistant depression is described as randomized, double-blind, active placebo-controlled, parallel-group ¹²⁸, and its CTIS-listed EPIsoDE protocol (NCT04670081) uses nicotinamide 100 mg as the placebo comparator against 5 mg and 25 mg psilocybin ⁷.

Expectancy and treatment-guess measurement. The field increasingly instruments the blind rather than only asserting it, capturing what participants and staff believe they received and how strongly they expected benefit. This practice is well established in adjacent psychiatric and psychotherapy trials (credibility/expectancy scales and patient/clinician treatment-guess forms) ⁷³⁷⁵, and FDA has now made expectancy and unblinding assessment an explicit expectation for MDMA specifically (see below) ²⁰⁰.

A telling detail is how openly sponsors concede the limits of these measures. Imperial College London's head-to-head trial of psilocybin versus escitalopram (NCT03429075) simply states that the blind is broken after six weeks, comparing 25 mg psilocybin plus placebo against 1 mg psilocybin plus escitalopram ¹.

Comparator arms

The registry record shows three comparator strategies, chosen largely by how much a sponsor prioritizes blind integrity versus a clean efficacy contrast.

• Inactive placebo. The pivotal MDMA Phase 3 trials (NCT03537014, NCT04077437) used an inactive placebo against MDMA, both delivered inside an identical manualized therapy course ^5153556069.
• Low-dose active control of the same drug. Used to preserve the blind, as in COMPASS P-TRD (1 mg vs 10/25 mg) ¹¹²⁰ and the Groningen CTIS study (1 mg vs 15/25 mg) ²⁵²²⁵³.
• Distinct active comparator. A newer MDMA PTSD trial (NCT05790239) abandons placebo entirely, using low-dose d-amphetamine-assisted therapy as an active comparator in a randomized, double-blind, two-arm design, an attempt to give the control arm its own noticeable drug effect ⁷². In depression, a NUDZ (National Institute of Mental Health, Czechia) study on the EU register compares psilocybin directly against ketamine ¹³⁰.

A structural feature cuts across all of these: the drug is embedded in psychotherapy, so the comparator is really "drug plus therapy" versus "control plus the same therapy." That confound is central to FDA's later critique.

Endpoints

Endpoint selection is comparatively standardized and mirrors the underlying indication.

• PTSD (MDMA). The primary endpoint across the MDMA program is change in the Clinician-Administered PTSD Scale for DSM-5 (CAPS-5) total severity score from baseline, measured at 18 weeks post-baseline in the pivotal studies and at 13 to 18 weeks in the open-label Phase 2 lead-ins ^{50576259606972}.
• Depression (psilocybin). The dominant primary endpoint is change in the Montgomery-Asberg Depression Rating Scale (MADRS). COMPASS P-TRD measures MADRS change to Week 3 ¹¹²⁰; the Groningen CTIS trial to 6 weeks after the second dose ^247252253; and Cybin's CTIS trial (2024-519270-40-00, 8 mg and 16 mg versus placebo) to Day 42 ^255256263. Not every protocol uses MADRS: the LKH Graz II psilocybin study (CTIS 2024-518654-16-01) uses the Beck Depression Inventory-II one week after four weeks of dosing ²⁶⁴.

The recurring difficulty is not which scale is chosen but two adjacent issues FDA has flagged: the short interval to the primary read-out, and the durability of any benefit beyond it.

The EU CTIS and EudraCT landscape

The European record is thinner and more exploratory than the US pivotal program. Under EU CTR (CTIS), psilocybin depression trials dominate, sponsored by a mix of academic centers and companies: UMC Groningen (2023-510488-36-01) ²⁵², GHU Paris (2024-512911-34-00) ²⁴⁹, Cybin IRL (2024-519270-40-00) ²⁵⁸, and LKH Graz II (2024-518654-16-01) ²⁶⁴. On the older EU Clinical Trials Register, the psilocybin footprint includes the Central Institute of Mental Health's active placebo-controlled Phase 2 trial ¹²⁸, Imperial College London ¹³², COMPASS Pathways' P-TRD and long-term follow-up records ¹³⁵¹³⁷, and the NUDZ psilocybin-versus-ketamine study ¹³⁰. MDMA appears mainly through MAPS Europe B.V. records for PTSD, including a long-term safety/persistence study and a feasibility study with an fMRI substudy ¹³¹¹³⁶. No LSD interventional trial surfaced in either European register in this review ¹²⁷.

Two data-quality caveats matter for anyone mining these registers. First, the CTIS and EudraCT phase fields are frequently inconsistent, with a title describing a Phase 2 study while the structured field records "Human pharmacology (Phase I)" ¹²⁸¹³⁶. Second, the public register rows often omit the primary-endpoint and comparator fields entirely, so design details have to be read out of the attached protocol rather than the metadata ¹²⁷¹³⁵.

What FDA raised at advisory committee and in the MDMA complete response letter

FDA took MDMA-assisted therapy for PTSD to the Psychopharmacologic Drugs Advisory Committee on June 4, 2024, where the product under review was MDMA capsules in support of an NDA for PTSD, and the committee was asked to weigh whether the data supported approval ^243266267. FDA then issued a complete response letter to Lykos Therapeutics for NDA 215455 (midomafetamine capsules) dated August 8, 2024, concluding that the application lacked substantial evidence of effectiveness and did not adequately establish safety ^199200201. The letter is unusually explicit about design, and it maps closely onto the blinding critique above.

Functional unblinding and expectancy bias. FDA noted that roughly 40% of enrolled participants had prior MDMA use, higher than expected for the PTSD population, raising selection bias and, critically, expectation and functional-unblinding bias concerns ²⁰⁰. In its path-forward recommendations, FDA asked future studies to actively assess participant expectancy and unblinding and to minimize enrollment of participants with prior MDMA or psychedelic use ²⁰⁰.

The psychotherapy confound. FDA questioned whether the psychotherapy component is even necessary and recommended that future development consider a factorial design that includes an arm with no psychotherapy, so the drug effect can be separated from the therapy effect ²⁰².

Durability of effect. FDA said the data did not show a durable treatment effect beyond the 18-week end-of-study assessment, and that the MPLONG follow-up was inadequate because it relied on a single visit with widely variable timing and selection-bias concerns ²⁰¹. The agency's recommended design fix was a randomized, double-blind study with blinded long-term follow-up and prespecified retreatment criteria, at least monthly follow-up visits, to establish durability ²⁰⁰.

Comparator and control design. Consistent with the dose-control approach seen elsewhere in the field, FDA suggested considering a low-dose midomafetamine arm as a control ¹⁹⁹²⁰⁰.

Safety, abuse potential, and data reliability. FDA said Lykos did not collect information on events that participants, therapists, or physicians considered "positive" or "favorable," which are relevant to abuse potential and impairment, and it identified unreported adverse events at study sites that undermined confidence in the safety data ²⁰¹. The agency cited inconsistencies between the MAPP2 protocol and the training/safety materials on what counted as an adverse event, contributing to under-reporting and inadequate characterization of acute effects, duration of impairment, and abuse-potential signals ²⁰¹. To address these, FDA recommended capturing all abuse-related adverse events, developing standardized discharge-readiness criteria, and considering an independent third-party audit of study records and treatment-session recordings ¹⁹⁹²⁰⁰.

Bottom line from FDA. The agency's stated view was that the most efficient path to approval would be a new clinical trial designed to demonstrate durability and adequately characterize safety, built around blinded post-treatment follow-up, prespecified retreatment, expectancy and unblinding assessment, a possible low-dose control, and reconsideration of whether psychotherapy is required ^199200201202. In effect, FDA asked Lykos to rebuild the trial around exactly the blinding, comparator, and endpoint-durability weaknesses that the registry record shows the whole field still struggles to solve.