For most of the last decade, the FDA’s reasoning on specific drug rejections has been opaque. Complete Response Letters, the formal documents the agency issues when it declines to approve a New Drug Application, are typically released only to the sponsor. Sponsors then choose how much to disclose publicly. The field, the press, and other developers in the same therapeutic class are left to infer the agency’s reasoning from secondhand accounts.

This changed on September 4, 2025. The FDA, under a transparency initiative, published the redacted full text of 89 previously unpublished Complete Response Letters issued in recent years. Among them was the August 9, 2024 CRL to Lykos Therapeutics, the document that formally rejected the MDMA-assisted therapy NDA. For the first time in the field’s history, the agency’s actual reasoning on a psychedelic submission is publicly readable.

This piece is a careful reading of that document. Every claim about what the CRL says is sourced to the public release; this is interpretation, not reporting on anonymous sources. The full document remains available through the FDA’s CRL release portal. Anyone who wants to verify any specific claim in this analysis can read the original alongside.

The piece is consequential beyond historical interest because the methodological positions documented in the CRL are positions the agency continues to hold. Sponsors of subsequent psychedelic programs (Compass, AtaiBeckley, Cybin, Usona, the Otsuka/Transcend methylone program) are operating in a regulatory environment where the standards documented in this letter are the standards that apply.

What a Complete Response Letter is, and what this one specifically says

A CRL is the FDA’s formal communication that an NDA is not approvable in its current form. It identifies specific deficiencies and typically requests specific additional work (additional clinical data, additional manufacturing information, additional safety analyses, additional commitments) that the agency would need to see in a resubmission. CRLs are not permanent rejections; they are statements of what the application would need to look like to receive approval.

The Lykos CRL identifies multiple categories of deficiency. The categories are reported here in the approximate order of prominence in the document.

The functional unblinding finding

The CRL documents the FDA’s analysis that nearly all participants in the active arm of the Phase 3 trials correctly identified their treatment assignment. This is the finding that received the most press coverage from the June 2024 advisory committee discussion, and the CRL formalizes it as a binding regulatory concern.

The letter’s framing is methodologically careful. The agency does not say that functional unblinding alone is fatal. Functional unblinding has been observed in many psychiatric trials, including the esketamine trials that supported Spravato’s approval. The agency’s concern is more specific: in the Lykos program, the functional unblinding combined with other design features in ways the agency could not adjudicate.

Specifically, the CRL documents that the trial design did not include sufficient mechanisms to disentangle drug effect from expectancy effect. The program did not use a dose-response design (it tested MDMA against placebo, not multiple MDMA doses against each other). It did not use an active comparator with similar somatic effects. It did not include pre-registered analysis of baseline expectancy as a predictor of response. The combination of functional unblinding with the absence of any of these expectancy-controlling design features left the agency unable to determine what fraction of the measured effect reflected the drug’s pharmacology.

This framing matters for subsequent programs. A psychedelic Phase 3 trial that has functional unblinding but also has dose-response separation, active comparator separation, or pre-registered expectancy analysis is not in the same regulatory position as the Lykos program was. The Compass COMP005 trial’s three-dose design produced dose-response separation; this is part of why the Compass program is not facing the same methodological objections.

The psychotherapy standardization finding

The CRL identifies serious concerns about the standardization of the integration therapy component across trial sites. The MDMA-assisted therapy protocol included multiple sessions of psychotherapy before, during, and after the dosing sessions, conducted by trained therapists.

The agency’s documented concerns include: the lack of a fully manualized protocol that could be replicated across sites, the variability in how individual therapists delivered the integration sessions, the absence of systematic fidelity monitoring across sites, and the methodological consequence that the agency could not separate the contribution of the drug from the contribution of the therapy or the therapist.

This is a structural issue specific to the Lykos program’s design, and it is one of the reasons the agency has been more comfortable with subsequent programs that incorporate drug-focused designs with more standardized psychotherapy components. The Compass program, for example, uses a more limited and more manualized psychological support framework. The methylone program uses a treatment paradigm that more closely resembles a conventional psychopharmacologic intervention. Both of these design choices are partial responses to the kind of concerns the CRL documents.

The safety reporting findings

This category of concern is the most operationally specific in the CRL and, in some ways, the most damaging to the program’s regulatory position.

The CRL documents that the trial design included instructions to clinical sites not to record certain euphoria-related effects as adverse events. The agency frames this as a problem because adverse event reporting in any clinical trial must be uniform, systematic, and not subject to selective non-reporting. A trial that systematically under-records certain effects creates a safety database that is structurally incomplete.

This is not a methodological subtlety; it is the kind of issue that makes a regulatory reviewer doubt the integrity of the entire safety dataset. The CRL’s framing suggests the agency could not separate adverse events that occurred but were not reported from adverse events that did not occur. The resulting uncertainty colored its read of the program’s safety profile.

For subsequent psychedelic programs, the implication is direct. Trial designs that attempt to define certain expected effects (intense subjective experiences, transient dissociation, altered states of consciousness) as not adverse events for purposes of trial reporting are likely to face the same kind of regulatory objection. The standard practice in the field has shifted, in the wake of the CRL, toward more comprehensive adverse event reporting protocols that capture the full subjective experience of dosing sessions and then categorize them analytically rather than selectively reporting them.

The prior drug exposure findings

The CRL documents that the trial included substantial numbers of participants with prior recreational MDMA use, and that this subgroup had different response patterns than participants who were MDMA-naive. The trial design did not pre-specify analyses of how prior exposure affected response, did not stratify enrollment to ensure adequate representation of both subgroups, and did not adequately characterize the population of participants by their pre-trial exposure history.

The agency’s concern is methodological. If response is different in MDMA-experienced versus MDMA-naive populations, the trial as conducted cannot tell the agency whether the drug works in the population that would actually be prescribed the drug commercially (predominantly naive patients with PTSD). The CRL specifically requests this analysis be conducted with subgroup stratification in any resubmission.

The implication for subsequent programs is that any psychedelic Phase 3 program needs to characterize and analyze its participant population by prior drug exposure, particularly for compounds that have meaningful recreational use patterns.

The durability findings

The CRL documents concerns about whether the program’s apparent benefits were durable enough to justify approval. The primary endpoint was measured at 18 weeks post-baseline, but the durability follow-up beyond that point was limited. The agency notes that the Phase 3 trials showed statistically significant improvements during the treatment period, but the long-term follow-up data was insufficient to conclude that the benefits persisted.

This is a recurring theme in the agency’s posture on psychedelic programs. The commercial case for psychedelic-assisted therapy depends substantially on the duration of effect from a small number of doses. If the effect duration is shorter than the commercial framing suggests, the value proposition narrows. The agency wants durability data that supports the commercial framing.

Subsequent programs have responded to this. Compass’s COMP006 included 26-week durability data as part of the registrational package. The methylone program’s IMPACT-1 study measured CAPS-5 at day 64 (approximately six weeks post-final-dose) and continues to track durability into ongoing follow-up.

What the CRL does not say

The CRL is also notable for what it does not contain.

It does not reject MDMA as a class of compound. The methodological concerns documented are program-specific, not mechanism-specific. A different MDMA program with different trial design could, in principle, address all of the agency’s concerns.

It does not reject the integration-therapy model in principle. The agency’s concerns about psychotherapy standardization are about implementation, not concept. A drug-plus-therapy program with a more rigorously manualized and fidelity-monitored psychotherapy component could, in principle, satisfy the agency.

It does not reject the use of psychedelic compounds in psychiatric indications. The CRL is internally consistent with subsequent agency actions: the breakthrough designations that have continued to be granted, the voucher selections in April 2026, the rolling NDA pathway granted to Compass. The agency is not rejecting the category; it is rejecting a specific application within the category.

The implications for the field

Reading the CRL as the field’s most detailed regulatory document on psychedelic methodology, three lessons emerge.

First, the agency is methodologically literate and specific in its concerns. Sponsors who present applications hoping that the agency will overlook trial design weaknesses are unlikely to succeed. The CRL demonstrates that the agency reads the trial designs carefully and identifies specific weaknesses, not as general dissatisfaction but as concrete methodological problems with concrete remediation paths.

Second, the agency’s standards apply uniformly across the field. The Lykos program received the same designations, the same breakthrough engagement, the same regulatory development pathway as the programs that subsequently received vouchers. The differential outcomes reflect differential methodological choices, not differential treatment by the agency.

Third, the CRL is now a public reference document, and subsequent submissions will be read against it. Any future psychedelic NDA that does not address the concerns documented in this CRL is operating without the benefit of one of the most specific roadmaps the agency has ever provided to a developing therapeutic class. Sponsors who have read it carefully are positioned differently than those who have not.

The September 2025 release of the CRL is, in some ways, the most consequential regulatory event in the psychedelic field’s recent history. Not because it announced any new policy, but because it made readable the agency’s actual reasoning on the most prominent rejection in the field’s history. Any sponsor planning a psychedelic Phase 3 program should treat the document as required reading. The Behavioral Wire archive will continue to track how subsequent programs respond to its specific concerns.