Definium’s Phase 3 LSD Win Is Real. The Fracture It Exposed Is Bigger.

Rachel Yehuda, Director of the Parsons Research Center for Psychedelic Healing at Icahn School of Medicine at Mount Sinai, put it plainly in a public conversation at Aspen Ideas: Health this week. “We can’t afford to lose the treatment for the molecule,” she said. “People who take psychedelics begin to feel things shift. But feeling changed is not the same thing as being healed.” She was not speaking in the abstract. She was speaking directly into the momentum generated by the most consequential psychedelic clinical trial result in history, and she was pumping the brakes.

Definium Therapeutics just delivered Phase 3 data that the field has been waiting two decades to see. A single 100 microgram oral dose of DT120, Definium’s LSD candidate, produced a placebo-adjusted MADRS difference of -8.1 points at six weeks (p<0.0001) in 149 adults with major depressive disorder, sustaining to -7.3 points at twelve weeks. Response rates hit 35% versus 7% on placebo. Remission reached 24% versus 3%. No serious adverse events. No suicidality signal. Per Tommaso Barba at Imperial College London’s Centre for Psychedelic Research, that 8.1-point separation is the strongest antidepressant effect recorded in any large-scale psychedelic trial to date, outpacing COMPASS Pathways’ Phase 3 psilocybin result at 3.8 points and Beckley Psytech’s 5-MeO-DMT Phase 2 signal at roughly 5 points. The stock surged. The company announced plans to raise $700 million in fresh capital. On the surface, this looks like a breakthrough moment.

But behind the fanfare lies a fracture that no Phase 3 p-value can close.

The Molecule Is Not the Mechanism

The field’s most credible voices are converging on a problem the FDA has no existing framework to resolve. Dr. Rick Barnett, a psychologist and ketamine-assisted psychotherapy clinician with over twenty years of practice, stated it without hedging: “The FDA approves drugs. It doesn’t approve whole health.” In a case-based review published in Primary Care Companion for CNS Disorders by clinicians at MGH and Harvard, he noted, prolonged adverse effects cluster in unguided settings, trust built during preparation predicts efficacy, and integration may require individualization for complex presentations. These findings are not fringe. They are mechanistic. And the FDA has no regulatory pathway for them.

What Yehuda and Barnett are identifying is a structural collision between how the FDA evaluates therapeutics and how psychedelic healing actually appears to work. The agency evaluates molecules against endpoints within defined windows. Psychedelic treatment, at least as the evidence currently frames it, is a process: preparation before the session, the session itself, and integration after it. Strip the process to isolate the molecule for regulatory review, and you may be approving something categorically different from what the trials tested.

Josh Hardman at Psychedelic Alpha reported this week that Definium’s own CMO Dan Karlin, along with external experts Sandeep Nayak and Martijn Arns, flagged several open questions around the Phase 3 result: the study’s size, the durability of effects, and how a Risk Evaluation and Mitigation Strategy (REMS) program would actually be structured. Nayak described the broader picture as “rapid and effective but less so than initial rosy perceptions, durable but not so much that redosing is not required.” Arns called the result “very exciting and encouraging” while noting that it fits a now-consistent pattern across psilocybin, 5-MeO-DMT, and LSD. Consistent pattern is not the same as solved problem. The REMS question alone could determine whether this drug is commercially viable or operationally trapped inside a delivery system that prices out the patients who need it most.

That delivery system question is where the field is quietly fracturing along ideological lines.

The Clinic Is Not the Only Container

Paul F. Austin, founder of Third Wave, surfaced research published in 2025 in Progress in Neuro-Psychopharmacology and Biological Psychiatry that deserves more attention than it has received. Eighty-five adults with histories of childhood trauma took psilocybin, ayahuasca, MDMA, or LSD in naturalistic settings: organized spiritual ceremonies or raves and electronic dance music festivals. Researchers assessed PTSD symptoms, complex PTSD symptoms, internalized shame, and sense of connection at three time points: before, within two days after, and two months later. Two months out, improvements were significant across every measure, on established clinical scales, regardless of whether participants had chosen ceremony or dance floor.

The study had no control group. Austin acknowledged that directly, and noted the reason: blinding a psychedelic trial is nearly impossible when participants can tell whether they received the active compound. This is the placebo problem Yehuda referenced at Aspen, the same problem HHS official Matthew Zorn described publicly as one of the two most pressing epistemological challenges in the field. The absence of a control group limits what can be concluded. It does not negate what was observed.

What it reveals is a tension the field has been slow to confront. If meaningful clinical-scale improvement in trauma symptoms occurs outside licensed clinics, in group ritual settings, without preparation delivered by a credentialed therapist and without integration structured by a licensed clinician, then the licensed-clinic model is not a treatment requirement. It may be a regulatory preference. And regulatory preferences, when embedded in REMS programs, become access barriers. Consider a scenario where DT120 reaches approval with a REMS requiring certified clinic administration and mandatory therapy components: the reimbursement infrastructure for that model does not exist at scale, the trained clinician workforce does not exist at scale, and the cost per patient will almost certainly replicate the access failure that followed SPRAVATO’s 2019 approval for treatment-resistant depression.

Barnett warned exactly this: “Reimbursement will follow approval. And approval will not include preparation. Will not include integration.” The Harvard and MGH clinicians publishing in Primary Care Companion for CNS Disorders documented why that matters mechanistically. These are not advocacy positions. They are clinical findings with regulatory consequences.

The Political Window and Its Limits

Layered beneath the science is a political moment that none of the researchers can fully control. Journalist Mattha Busby reported for the Evening Standard this week on Bryan Hubbard, a Republican attorney from Tennessee who leveraged a relationship with Joe Rogan to arrange an Oval Office meeting with President Trump, resulting in an executive order accelerating research into psychedelic therapies. Hubbard’s focus is ibogaine, specifically its documented effects on veterans with PTSD and traumatic brain injury, and early research suggesting it promotes new growth proteins in the brain with a durability that other psychedelics have not yet demonstrated. Hubbard helped persuade more than half a dozen states to fund ibogaine development trials before the federal executive order was signed. That is real political traction, built outside the traditional sponsor-FDA pathway.

But political traction and regulatory infrastructure are different animals. An executive order accelerating research does not resolve the blinding problem, does not create a reimbursement code for integration therapy, and does not answer the question Yehuda posed publicly this week: what is the clinical context in which psychedelics should be administered, and who is responsible for ensuring that context exists after approval? The FDA’s current framework was built for molecules with defined administration routes and measurable blood levels. It was not built for a treatment where, per the evidence, what happens in the room before and after the dose may be as consequential as the dose itself.

Definium’s MADRS numbers are real. The p-values are not in dispute. What is in dispute is whether those numbers, generated in a controlled clinical trial with therapy protocols, will replicate in the licensed clinic down the road where the psychiatrist has thirty minutes, the integration therapist is not reimbursed, and the patient spent the session alone on a recliner with noise-canceling headphones. That gap between controlled trial and clinical reality is where every promising psychiatric drug has gone to underperform for the last thirty years. Psychedelics have better Phase 3 numbers than most. They do not yet have a better implementation model. The field has until FDA filing to build one, and based on this week’s conversation, the people who understand the problem most clearly are the ones least likely to be in the room when the REMS is written.

References

1. Tommaso Barba, PhD candidate, Imperial College London Centre for Psychedelic Research. LinkedIn post on Definium Phase 3 LSD results. 2025.
2. Paul F. Austin, Founder, Third Wave. LinkedIn post on clinic settings and naturalistic psychedelic healing research. 2025.
3. Josh Hardman, Founder, Psychedelic Alpha. LinkedIn post on expert reactions to Definium Phase 3 data. 2025.
4. Rachel Yehuda, Director, Parsons Research Center for Psychedelic Healing, Icahn School of Medicine at Mount Sinai. LinkedIn post on Aspen Ideas: Health conversation with HHS. 2025.
5. Dr. Rick Barnett, Co-Founder, Psychedelic Society of Vermont; Owner, BPSHealth. LinkedIn post on FDA drug approval and psychotherapeutic context. 2025.
6. Mattha Busby, Journalist. LinkedIn post on Bryan Hubbard and the Republican psychedelic policy movement. Evening Standard. 2025.