TS-161 was supposed to work like ketamine and behave like a normal drug: no dissociation, no abuse risk, no need for a monitored infusion clinic. A small trial run at the NIH tested that idea directly in patients for the first time, and it failed. The drug showed no advantage over placebo on the primary measure of depression severity, no participant reached remission on either arm, and secondary measures showed nothing either. That result matters less on its own than for what it joins: a growing list of drugs built on the same rational premise, replicate ketamine’s mechanism, drop its baggage, that keep failing the same test.
What the trial found
Eleven adults with treatment-resistant depression took TS-161 or a placebo for three weeks, then switched to the other arm. The study was supposed to enroll 25 people but was stopped early when recruitment ran too slow. At the primary endpoint, day 21, there was no significant difference between drug and placebo, a result about as flat as a trial can produce. One person out of eleven responded to the drug by day seven; zero of nine responded to placebo at the same point, too small a gap to mean much on its own. Nobody in the trial reached remission.
What did move were three biological measurements that had nothing to do with how patients felt. Brain recordings showed increased gamma-frequency activity on the drug, matching a pattern seen with ketamine itself. A blood marker linked to brain plasticity trended upward, though not by a statistically meaningful amount. Brain chemistry scans suggested a hint of increased glutamate, a qualitative trend the trial was too small to test formally. The drug appears to have done something to the brain. It didn’t make anyone better.
Why this drug exists at all
Ketamine works fast and helps people who’ve failed every standard antidepressant, but it also causes short-term dissociation and carries real abuse potential, which limits it to monitored clinics and keeps it from becoming an everyday pill. TS-161 targets a different molecular switch, a receptor called mGlu2/3, that lab studies suggested could produce ketamine-like effects on mood circuitry without the trip or the addiction risk. That’s a real, scientifically reasonable idea, and the reasoning stated across this research group’s own work is plain: this class of drug engages many of the same cellular pathways as ketamine without being expected to carry ketamine’s side effects or abuse potential.
This isn’t the first time that idea has failed
An earlier drug in the same receptor class, decoglurant, was tested in a 357-patient placebo-controlled trial for depression and showed no benefit on any measure, a result published in 2020. And a completely different approach aimed at the same goal, esmethadone, an opioid-derived compound with no meaningful opioid activity at treatment doses, designed to preferentially block the overactive form of ketamine’s target receptor while leaving normal brain signaling alone, missed its own primary endpoint in a 227-patient Phase 3 trial in 2022. Post-hoc analyses published over the past year found encouraging signals in specific subgroups, patients with severe depression and those who’d relapsed after an initial antidepressant response. Its maker, Relmada Therapeutics, walked away anyway: after a second Phase 3 was declared futile at an interim analysis in late 2024, the company discontinued the program, gave up its license to the drug in 2025, and left depression development entirely. No esmethadone trial is running anywhere today.
Three drugs. Two different molecular targets. The same basic design idea. The same result on the test that was supposed to prove it works.
What this tells you, and what it doesn’t
None of this proves ketamine’s mechanism can’t be replicated safely. Every one of these trials had a real limitation working against it: TS-161’s trial enrolled fewer than half its planned patients, esmethadone showed post-hoc signals in specific patient subgroups even after its main test came up short, and drug development regularly takes multiple failed attempts before someone gets the dose, the population, or the trial design right. What this pattern does show is that the scientific logic connecting a lab mechanism to a clinical result is not, on its own, a reliable predictor of what happens in actual patients. TS-161 engaged the brain in ways that looked like ketamine on a scan. It did not act like ketamine in a person.
What would change this picture
The fix future trials would need follows from the design itself: larger patient groups, and ideally a way to select patients based on their biology beforehand, instead of testing every treatment-resistant patient and hoping the mechanism helps broadly. If a biomarker like the gamma-wave signal seen here could identify, before treatment starts, which patients are likely to respond, that would turn this from a failed drug into a partially validated concept still waiting for the right trial. Nothing published so far does that yet. Until it does, this specific approach to building a better ketamine remains a good idea without a supporting result.