"There are drugs in the drug development graveyard that probably shouldn't have been there — that probably could have benefited patients. We just didn't have the right tools to demonstrate what needed to be demonstrated."
— Matthew Leoni
Matthew Leoni didn't set out to be a drug developer. He went to Penn for medicine, added an MBA, and was on track to practice pathology. Then he looked at what that life would actually look like — and made the turn. Nearly two decades later, he's built a career spanning derm, immunology, MS, schizophrenia, and Parkinson's, most recently as Senior VP of Global Clinical Development at Cerevel before its acquisition by AbbVie. Now he's building something new as CMO of Merida Biosciences.
In this episode, Brandon Li sits down with Matt to work through the real constraints of neuro drug development — why trials fail even when molecules work, how the industry has gotten locked into endpoints it can't escape, and where the genuine opportunities lie in the decade ahead.
Why Neuro Is the Hardest Therapeutic Area
Matt doesn't soften it: neuro has the highest failure rates of any therapeutic area, and the reasons are structural. Most endpoints in neuropsychiatry are subjective — clinician interpretations of how a patient feels or behaves on a given day. That subjectivity produces variance. Variance collapses trials.
Placebo response compounds the problem. In major depressive disorder, schizophrenia, and related conditions, the placebo effect is not just real — it's growing. Year over year, the historical metadata shows placebo response rates trending upward, making drug-placebo separation harder to achieve in controlled settings. Whether it's patient population saturation, recycled participants, or something else, the trend is clear.
"Twenty-five years ago there weren't all these ongoing trials. The patient populations were fresher. Whatever the reason, year over year it just keeps getting harder to show drug effects in an experimental design."
The Endpoint Trap
One of the conversation's sharpest moments is Matt's description of what he calls retrofitting — when drug developers don't choose endpoints that best measure their molecule, but instead shape their programs to fit what regulators have already accepted.
The PANSS (Positive and Negative Syndrome Scale) in schizophrenia is the clearest example. It's decades old. It has real limitations. But it's accepted. So every sponsor runs their drug against it, not because it's the optimal tool, but because they know it works with the FDA. Innovation in measurement — EEG, facial expression analysis, wearable biometrics — sits on the sidelines, too risky to use in a pivotal trial, too slow to develop as a proprietary endpoint.
"You're basically incentivized to use what works already. The only time you get pushed off the reservation is when there's no precedent — when you have no choice."
The result: drugs that might have worked under the right measures get one shot in a pivotal trial, fail to separate, and go into the graveyard. Whether the mechanism was wrong, or just the tool for detecting it, is rarely knowable afterward.
Disease Modification: Where the Field Needs to Go
Most of Matt's career has been on the symptomatic side — finding better ways to treat the manifestations of disease without changing its trajectory. He's candid about the limits of that work: in schizophrenia, in major depressive disorder, in Parkinson's, we're still fundamentally playing with neurotransmitters rather than interrupting what's actually causing disease progression.
Disease modification — the idea of slowing or reversing the underlying pathology rather than just managing symptoms — is the field's north star. And Matt believes the tools to get there are finally arriving: AI-assisted target identification, CRISPR, improved understanding of neural network complexity. The oncology analogy is instructive. Chemotherapy was the broad sword. Immunotherapy was the inflection point. Neuro is still waiting for its inflection.
"We're on the verge of understanding the neural networks well enough to start producing targets that can have real meaningful impact. I'm hoping we see a couple more wins before a couple of big hits — because it's at a tenuous spot right now."
On Psychedelics and Psychoplastogens
Matt watches the psychedelic space with interest and some caution. The regulatory challenges are real — MDMA's FDA setback in particular reflects how hard it is to design trials for compounds where true blinding is effectively impossible. But he's more optimistic about a different category: psychoplastogens — compounds like those being developed at Delix — which aim to capture the neuroplasticity benefits of psychedelics without the hallucinogenic experience.
"We're starting to burgeon on the concept of disease modification through neuroremodeling. You're not just interdicting in a receptor — you're remodeling neurons. That's genuinely exciting."
Whether that excitement translates into approved medicines will depend on the same thing it always does in neuro: building the right tools to measure what the drug is actually doing.
