Seltorexant is a new kind of antidepressant. It doesn’t touch serotonin or dopamine directly. Instead, it blocks a receptor called OX2R, part of a brain system called orexin that controls arousal and sleep. Janssen has been testing it for years as an add-on treatment for people with depression who also have insomnia, a combination current antidepressants handle poorly. A detailed review of the drug’s evidence, published August 29 by an Italian research group led by Marco Di Nicola, lays out everything seltorexant has going for it. It also contains the one number that matters most for whether the drug ever reaches patients, and that number is a loss.
A drug built to replace quetiapine, tested against quetiapine, and it lost
Quetiapine is the drug doctors reach for when a depressed patient also can’t sleep, usually at a low, off-label dose. It works reasonably well but carries serious risks: weight gain, metabolic problems, a higher long-term cardiovascular risk. Seltorexant’s entire commercial case rests on offering the same benefit without those risks. So Janssen ran the test that matters most: a head-to-head trial, called MDD3005, pitting seltorexant directly against quetiapine over 26 weeks in patients with depression and significant insomnia. Seltorexant needed to either match or beat quetiapine’s response rate, the share of patients whose symptoms improved by a set amount. It did not. The trial’s primary endpoint, response rate at 26 weeks, was not met. The review reports seltorexant showed a numerically higher rate than quetiapine, but the difference was too small to rule out chance.
What still worked, and why it isn’t enough on its own
Seltorexant’s trial data isn’t weak. In its main placebo-controlled Phase 3 trial, called MDD3001, patients on seltorexant improved more than patients on placebo, a statistically solid result, though a modest one: a difference of 2.6 points on a standard 60-point depression scale. A separate combined analysis of five trials across this drug class found seltorexant at 20 milligrams was the only version, dose or compound, that showed a measurable antidepressant effect at all. A related orexin-targeting compound in Janssen’s pipeline, acting on a different receptor subtype than seltorexant, failed to separate from placebo at all. That’s an interesting scientific finding on its own: something specific about blocking this exact receptor, not just sedating a patient into feeling calmer, appears to be driving the effect. But a drug doesn’t get prescribed because its mechanism is interesting. It gets prescribed because it beats, or at least matches, what a doctor would otherwise use. Against quetiapine specifically, seltorexant hasn’t shown that yet.
The side of the story the review’s own optimism can’t fully cover
To their credit, the paper’s authors don’t hide the head-to-head loss. They discuss it directly and offer several reasons it might have happened: quetiapine is already a strong competitor, a hard drug to beat on a single yes-or-no measure; the specific measure used, response rate, is a blunter tool than the continuous symptom scores where seltorexant looked better; and the trial might not have enrolled enough patients to detect a modest advantage even if one exists. All of that is plausible. None of it changes the result itself. The authors’ proposed fallback position, that seltorexant might still win on safety and tolerability even without beating quetiapine on effectiveness, is a defensible argument. It’s also a smaller, harder sell to a prescriber than “this drug works better,” which is what the head-to-head trial was designed to test.
Gaps the authors name themselves
Two specific symptoms of depression, low motivation and pleasure (called anhedonia) and suicidal thinking, are exactly the symptoms clinicians most need a new drug to address. The review states plainly that no published data exists breaking out seltorexant’s effect on either one specifically. Long-term safety data beyond the trials already run doesn’t exist yet either. None of these gaps are unusual for a drug still in trials. They’re worth naming because a company’s own review of its drug choosing to flag them directly is a stronger signal than an outside critic making the same point.
What decides seltorexant’s fate from here
A third Phase 3 trial, called MDD3003, is still running and expected to reach its primary completion by the end of this year. It’s a longer trial built to test how durable seltorexant’s effect is over time, and the review calls it the single most important remaining piece of evidence. A strong result there, paired with the already-positive placebo trial, could still support an approval application. A mixed result would likely mean Janssen needs to run yet another large trial before deciding whether to file at all, pushing any possible approval years further out. Either way, the head-to-head loss against quetiapine doesn’t disappear. Even a drug that eventually gets approved still has to answer the question that trial was built to settle: not whether it works, but whether it works better than what’s already sitting on the shelf.