An LSD pill posts the best depression data the field has seen — and the hardest result to read

Biotech & Longevity 5 min 80 sources

Definium's psychedelic drug beat placebo by 8.1 points in a phase 3 trial and sent its shares up 55%. The number is real; the trouble is that a drug which makes you trip can't be tested blind, so no one can fully separate the drug from the patient knowing they got it.

Key takeaways

An LSD-based pill from Definium beat placebo by 8.1 points in a phase 3 depression trial — but because psychedelics make patients know they got the drug, no one can cleanly separate the medicine from the patient's expectation.
A single Alzheimer's patient reportedly improving after psilocybin is a reason to run a trial, not proof of a cure — it's one uncontrolled case.
Around the headlines, the field ran as usual: Merck and Otsuka won phase 3 trials, Pfizer's lung-cancer drug failed, regulators moved to speed early trials and to weigh loosening peptide rules, and the US sent an experimental treatment into Congo's deadly Ebola outbreak.

This week’s biggest biotech story is also its trickiest to judge. A psychedelic drug delivered numbers a depression researcher would dream of — and the way psychedelics work makes those numbers unusually hard to trust. The rest of the week was a normal run of the machine: two trials won, one failed, and regulators moved on two fronts at once.

The headline: a strong result you can’t fully see

Definium Therapeutics said a single dose of its LSD-based pill, DT120, sharply cut depression symptoms in a late-stage trial. Its shares jumped 55% to a four-year high [1][2]. The drug is a pharmaceutical version of lysergide — LSD — and works by switching on serotonin receptors in the brain [1].

The number was the story. In the 149-patient Emerge trial, DT120 beat a placebo by 8.1 points on the MADRS, the standard depression rating scale doctors use to score severity [3][1]. One Wall Street analyst called the result “unprecedented in the depression space” [1]. The company’s own CEO — at a firm that until January was called MindMed — said it was “the best data ever seen in a pivotal study of depression” [3]. Definium reported no new safety signals and no rise in suicidal thinking [3].

A phase 3 trial is the big, final test before a regulator decides on approval — hundreds of patients, the drug measured against a dummy pill or the current standard [3]. So this is a serious result, not a press-release tease.

Here is the caveat the headline numbers can’t carry. You cannot run a fair blind test of a drug that makes you hallucinate. In a normal trial, neither the patient nor the doctor knows who got the real drug and who got the placebo — that’s what keeps expectation from doing the work. But a psychedelic announces itself: the patient who gets DT120 knows they got it, and the patient who got a sugar pill knows they didn’t [3]. Researchers call this “functional unblinding.” When the people in your treatment group know they’re being treated, some of that 8.1-point gap is the drug, and some is hope, and the trial can’t cleanly tell you how much is which. Definium’s own slides note you can’t directly compare its number to rival trials because the methods and patients differ [3]. The result is genuinely strong. It is also genuinely hard to read — and both of those are true at once.

The same week, a single patient and a magic mushroom

The blurry line between drug and expectation showed up in a much smaller, much louder story. A Brazilian group reported that an 83-year-old woman with severe Alzheimer’s — who hadn’t spoken more than single syllables in years — began holding conversations after a large dose of psilocybin, the compound in magic mushrooms [4]. She reportedly regained some bladder control and mobility too [4].

It reads like a miracle, which is exactly why the caveat matters. This is a single, unblinded, uncontrolled case report — one person, no comparison group, observed by an organisation that practises psychedelic medicine [4]. The researchers themselves stressed her Alzheimer’s was not reversed; they suggest the drug may have tapped residual function rather than restored lost neurons [4]. Psilocybin, like LSD, hits serotonin receptors and is thought to briefly boost the brain’s plasticity [4]. One striking person is a reason to run a real trial — it is not yet a result.

Trials: two hits, one miss

The week’s ordinary trial news ran the usual three directions.

Merck said tulisokibart, an antibody it acquired in an $11 billion buyout in 2023, met its main goal in a phase 3 ulcerative colitis trial — the first drug of its kind to push patients into remission at 12 weeks, the company said, with no safety concerns flagged [5]. Otsuka posted a phase 3b win for its ADHD drug centanafadine in 315 adults who also had anxiety, weeks ahead of a possible US launch [6].

And one missed. Pfizer’s first new antibody-drug conjugate from its Seagen acquisition — a drug that ferries a toxic payload straight to cancer cells — failed its phase 3 lung cancer trial [7]. That’s the field working as designed: most drugs that reach this stage still fail, and the failures rarely make the front page.

Regulators move on two fronts

US health agencies pushed on access from opposite ends this week. They unveiled “Operation Trial Blazer,” a plan to speed up early-stage phase 1 trials by 6 to 12 months and pull them back from China and Australia, where they’ve been drifting [8]. Phase 1 is the first human test, checking mainly that a drug is safe.

At the same time, an FDA advisory committee said it will meet in July to discuss easing restrictions on peptides — short chains of amino acids that include real blockbusters like the GLP-1 weight-loss drugs, but also a thriving online gray market of “research” injectables of dubious quality, sold with thin evidence and heavy influencer promotion [9]. Loosening the rules would let US compounding pharmacies legally fill them [9].

The under-covered story: Ebola in Congo

While the markets watched a psychedelic stock, the US quietly sent an experimental Ebola treatment to the Democratic Republic of Congo, where an outbreak of the Bundibugyo strain has caused more than 1,000 cases and over 250 deaths [10]. There is no approved vaccine or treatment for this strain [10]. The drug is going out for compassionate use and to seed a clinical trial in the outbreak zone — the rare case where the emergency and the experiment are the same event, and the data gathered now could inform a future approval [10].

02 · Lesson · why it matters

Why some answers hide the moment you try to measure them

The fairest test keeps the subject from knowing which group they're in — but some treatments announce themselves, and then hope and the drug get tangled past separating.

The trick that makes medicine trustworthy

For most of history, a doctor’s belief in a remedy was its main evidence. People got better; the doctor took credit. The trouble is that people often get better anyway — from time, from rest, from believing they were being helped.

So medicine learned a trick. Split the patients in two. Give one group the drug, the other a fake. And — this is the clever part — make sure no one knows who got which. Not the patient, not even the doctor scoring them. That’s a blind trial. Hide which group people are in, and you strip out the one thing that always lies: expectation.

When Definium’s drug beat a sugar pill by 8.1 points, that gap is supposed to be the drug alone. The blinding is what makes the gap mean something.

What happens when the treatment tells on itself

Now give someone LSD. They will know. The room breathes, colours move, the hours bend. The patient who got the real drug knows they got it. The patient who got the sugar pill knows, within an hour, that they got nothing.

The blind is broken — not by a leak, but by the drug doing exactly what it does. Researchers have a flat name for it: functional unblinding. Once both groups know which arm they’re in, expectation walks back into the room. The treated patient hopes. The placebo patient sinks, knowing they drew the short straw. The gap between them grows — and now it’s part drug, part hope, part disappointment, and no statistic can pull those threads apart.

This isn’t a flaw in one trial. It’s a fact about a whole class of treatments. Anything you can feel working — a psychedelic, a stimulant, a drug with a sharp side effect — partly tells the subject the answer. The measurement reveals itself to the thing being measured, and the cleanest tool we have goes blurry.

The pattern is everywhere we test people

Step back from the clinic and you find the same shape wherever we try to measure an effect on people who can tell they’re being measured.

Put a new teaching method in one classroom and the old one in another, and the teachers running the new one try harder — they know they’re the experiment. Test whether a workplace policy lifts morale, and the team that knows it’s the pilot behaves like a team being watched. Measure whether a diet works, and the person who chose it wants it to work, and reports accordingly.

In each case the subject can read which group they’re in, and that knowledge changes what you’re trying to measure. The honest answer is hiding behind the act of looking for it. We don’t usually call it functional unblinding outside medicine, but it’s the same beast: you cannot keep a test fair from a participant who can see the test.

We are the worst-blinded subject of all

Here is where it reaches you, and reaches the people who run the trials too. The hardest experiment to keep honest is the one you run on yourself.

You start a supplement, a new routine, a way of eating, and you watch for it to work — because you chose it, paid for it, hoped for it. You are the patient who knows they got the real drug. Every good week feels like proof; you forget the good weeks you had before. You are not lying. You simply can’t blind yourself to your own hope, and your hope is a real effect, sometimes a large one.

This is why “it worked for me” is the weakest evidence there is, and also the most convincing — to the person who lived it. The woman who spoke again after psilocybin is real, and her son’s amazement is real. One unblinded life is a reason to look closer. It is not yet an answer.

So when a stunning result arrives — the best the field has seen — hold two things at once. The number may be genuine. And the cleaner the feeling, the harder it is to know how much was the thing and how much was the wanting. The people inside the trial can’t fully see that. Neither can the analyst, the company, or you. None of us stands outside our own hope long enough to measure it — and knowing that is the closest any of us gets to standing outside it.

03 · Lab · your turn

Read the Trial

Rehearse splitting a drug's measured result into real effect and the patient's expectation, and feel how a treatment you can feel breaks a blind test.

04 · Hope · carry this

It is a quiet triumph that we ever built a way to catch ourselves hoping — the blind trial exists because people were honest enough to admit they couldn't trust their own eyes, and then clever enough to do something about it.