The FDA Rejected MDMA Therapy Because It Couldn't Trust the Data. The New Application Doesn't Have New Data.

On November 22, 2014, eight years after he came home from Iraq with what he calls “crippling” PTSD and five suicide attempts behind him, Jonathan Lubecky took his first dose of MDMA inside a clinical trial in Charleston. He’d found the trial during a psychiatric hospitalization, the kind of last-stop referral that happens after every other approach has already failed a person. His PTSD score dropped by roughly half on the standard clinical scale. A decade later, testifying before the House Veterans’ Affairs Committee — the first congressional hearing ever held on the subject — he put what that trial meant to him in eleven words: “MDMA-assisted therapy is why my son has a father and not a folded flag.”

That trial, and the ones like it, are the exact data the FDA looked at in August 2024 and said it could not trust.

This month, the company that ran those trials — renamed Resilient Pharmaceuticals after its predecessor, Lykos Therapeutics, laid off most of its staff following the rejection — resubmitted MDMA-assisted therapy for PTSD to the FDA under the proposed brand name Rysanso. It did not run a new efficacy trial. It submitted a third-party audit of the old trials, a new safety study on cardiac effects, and follow-up data gathered from VA patients. The clock on the roughly six-month review started the day the application landed.

The FDA’s objection in 2024 was never that MDMA doesn’t work. It was that nobody could prove it works blind — and the new application still can’t.

The problem was never the drug. It was the study design.

Here’s the layman version of what killed the 2024 application, because the jargon the FDA used — “functional unblinding” — is doing a lot of work to hide something simple. A blind clinical trial depends on nobody knowing who got the real drug and who got the placebo. MDMA makes that nearly impossible. Take a real dose and you feel it — the warmth, the openness, the unmistakable shift in your own head. Take a sugar pill and you don’t. So the patient knows within the first hour which group they’re in, and so does the therapist sitting across from them, and once both of them know, every rating either of them gives afterward is contaminated by that knowledge. An FDA advisory committee voted 9-2 that the trials hadn’t established effectiveness and 10-1 against the drug’s benefit-risk balance — not because the therapy didn’t seem to help the people in the room, but because the way the trial was built made it structurally impossible to separate “this drug works” from “everyone in this room knew who got the drug and behaved accordingly.”

That is not a small methodological quibble. It’s the difference between a therapy that has been proven and a therapy that has been believed, sincerely, by people who wanted it to work — which describes almost every patient, therapist, and advocate involved in the psychedelic-therapy movement, this reporter included in the sense that the evidence for MDMA’s promise is genuinely compelling. Wanting something to be true and it being provably true are different facts, and the FDA’s entire job is refusing to conflate them.

What makes this month’s resubmission strange is that it doesn’t resolve that problem. It works around it. Resilient’s application leans on a third-party audit of the original trial data — checking the numbers that already existed, not generating new ones — plus a phase 1 study confirming MDMA doesn’t pose unacceptable cardiac risk, plus longer-term follow-up from the veterans who went through VA-adjacent versions of the therapy. None of that addresses the actual finding that took the application down in 2024. An audit can confirm the original data was recorded honestly. It cannot make the original data un-unblinded.

The FDA wrote the rulebook after the game had already been called

There’s a specific irony sitting on top of this timeline, and it’s worth naming directly. On July 14, 2026 — after the 2024 rejection, and while Resilient’s resubmission was still being assembled — the FDA finalized formal guidance telling psychedelic drug sponsors exactly how to solve the unblinding problem going forward: use an active comparator instead of an inert placebo, build in blinding questionnaires, use raters who don’t know which arm a patient is in, measure participants’ own confidence about which group they think they’re in and control for it statistically. It’s a genuinely useful document. It is also, for Resilient’s specific application, a rulebook published after the test it was meant to fix had already been graded and returned. The company isn’t running a trial designed to this new standard. It’s asking the FDA to approve a drug using trials that predate the standard, on the theory that an audit plus real-world follow-up can stand in for the redesign the guidance describes.

It is also, for Resilient’s specific application, a rulebook published after the test it was meant to fix had already been graded and returned.

Meanwhile, the trial that could actually generate the clean data everyone agrees is missing is already running — just not for the company that needs it fastest. The Department of Veterans Affairs began enrolling roughly 80 veterans in June 2026 in a randomized, placebo-controlled trial of MDMA-assisted therapy for treatment-resistant PTSD with co-occurring alcohol use disorder, led by Dr. Erica Eaton at the Providence VA Medical Center with a second site in Connecticut. It’s designed with an active placebo, the exact fix the FDA’s new guidance recommends. Results aren’t expected until 2030. The veterans enrolling in it right now are, in effect, building the evidence base that Resilient’s application is trying to bypass — on a timeline four years slower than the company’s own review clock.

This isn’t the first time Washington decided the evidence didn’t matter

If the shape of this fight feels familiar, it should. On July 1, 1985, the DEA placed MDMA into Schedule I on an emergency basis — the first time the agency had ever used that emergency power on any drug. At the time, MDMA was being used, legally, by a small community of psychiatrists who found it useful for couples therapy and trauma work; four of them testified in the DEA’s own hearings that it had an accepted medical use. The DEA’s own administrative law judge, Francis Young, spent a year hearing the evidence and recommended Schedule III — restricted, but medically available. The DEA administrator overruled his own judge and scheduled it as Schedule I anyway, arguing that the lack of FDA approval alone was enough to justify the harshest category regardless of what the evidence showed. Three years later, a federal court agreed with Young’s original reasoning and ordered the DEA to reconsider. The DEA kept MDMA in Schedule I anyway. That was 1988. It is still there.

The pattern isn’t that regulators are wrong to demand rigor — Judge Young’s hearings and the FDA’s 2024 rejection both involved real, defensible scientific reasoning. The pattern is what happens in the gap between what the evidence shows and what the agency in charge decides to do about it, and how long ordinary people wait inside that gap while it gets sorted out. In 1985, the wait was measured in decades of Schedule I research restriction. In 2026, it’s measured in a veteran enrolling in a trial that reports results in 2030, while a company’s faster-track application asks the same agency to trust an audit instead.

If you’ve sat in the circle, you already know what a p-value can’t hold

If you have ever sat in a room — a VA group, a peer-support circle, a kitchen table at 2 a.m. — and listened to someone describe what actually happened for them in a therapy session, you already know something a clinical trial structurally cannot capture: that the “unblinding” the FDA is worried about isn’t a flaw in the therapy, it’s the therapy. Knowing you took the real drug, feeling the shift, trusting the person sitting across from you enough to go somewhere you couldn’t go sober — that’s not contamination of the data. For the person in the room, that’s the whole mechanism. The FDA’s caution and the patient’s experience aren’t actually in conflict; they’re answering two different questions. One is “did this help you.” The other is “can we prove, to a stranger who wasn’t in the room, that it wasn’t just the feeling of being cared for by someone you trusted.” Both questions are legitimate. Only one of them is currently unanswered in a way the law requires before insurance can pay for it.

If you’re a case manager or clinician with a veteran or first-responder client asking about MDMA therapy this week, the useful thing to tell them isn’t “it’s about to be approved” — it isn’t, not on any clock shorter than the FDA’s review window, and that window has failed this exact drug once already. The useful thing is to point them toward what’s real right now: the VA’s own psychedelic research program, now backed by more than $23 million in external grants across 19 active studies, is actively enrolling in multiple cities, and legal ketamine-assisted therapy — a different drug, but the closest currently-approved analog for supervised psychedelic-assisted trauma work — is available today at licensed clinics without waiting on any FDA decision at all. Neither is MDMA. Both are real, fundable, and reachable this month, which is more than the resubmission can currently promise.

Both are real, fundable, and reachable this month, which is more than the resubmission can currently promise.

Jonathan Lubecky’s trial happened in 2014. He testified about it in 2023. The company that ran it is now three names and one rejection removed from where it started, betting a rulebook written after the fact can substitute for the trial it never re-ran. Somewhere between his son growing up with a father and a veteran enrolling in a study that won’t report until 2030, the FDA is going to have to decide, again, what counts as proof — and how many more people are supposed to wait through the sequel while it does.