Seltorexant

Seltorexant (development code JNJ-42847922) is a synthetic small molecule that blocks the orexin 2 receptor (OX2R), one of two receptors for the arousal-related orexin peptides (Yun et al., 2017). Janssen scientists characterized it in 2015 as a selective OX2R antagonist and clinical candidate for insomnia (Bonaventure et al., 2015). It was later tested in major depressive disorder, and a 2025 review called it investigational (Thase, 2025).

Published work on seltorexant falls into six areas, each with its model named. Every human trial cited here has Janssen or Johnson & Johnson authors.

Discovery as JNJ-42847922

Janssen chemists optimized a series of octahydropyrrolo[3,4-c]pyrroles as selective orexin 2 antagonists that promoted sleep in rats; compound 34, JNJ-42847922, became the clinical candidate for primary insomnia (Letavic et al., 2015). A companion paper reported high OX2R affinity, about 100-fold selectivity over the human orexin 1 receptor, and fast brain OX2R occupancy and clearance after oral delivery in rats (Bonaventure et al., 2015). Models: binding assays and rats.

Seltorexant mechanism of action at the orexin 2 receptor

Rats given 3 to 30 mg/kg by mouth in the light phase reached non-rapid eye movement (NREM) sleep sooner and spent more time in NREM sleep over the first 2 hours, with little change in REM sleep; the effect held over 7 days at 30 mg/kg. It promoted sleep in wild-type but not OX2R knockout mice, consistent with an OX2R-mediated effect, and unlike zolpidem it raised neither nucleus accumbens dopamine release in rats nor place preference in mice (Bonaventure et al., 2015). Models: rats and mice.

Stress hormones and anxiety-like behavior in mice

Cage-exchange stress raised serum ACTH in mice. Pretreatment with JNJ-42847922 (30 mg/kg by mouth) abolished the rise, while the dual orexin antagonist SB-649868 only partly reduced it; the authors judged the ACTH effect separate from sleep induction (Yun et al., 2017). An independent group found that seltorexant given intraperitoneally increased baseline anxiety-like behavior in male mice, while the OX2R agonist YNT-185 reduced it (Chen et al., 2025). Models: mice.

Seltorexant insomnia and sleep studies

In healthy volunteers given ascending daily amounts or placebo for 10 days, seltorexant was absorbed rapidly (median time to peak 0.5 to 1.5 hours, mean half-life 2 to 3 hours) and induced somnolence at higher amounts, with no clinically relevant change in other central measures beyond a mild drop in attention (van der Ark et al., 2018).

In a phase 2 crossover study in insomnia disorder, polysomnography showed sleep efficiency 5.8 and 7.9 percentage points above placebo on nights 1 and 5 (both p < 0.001), with longer total sleep and shorter latency to persistent sleep among 27 completers (De Boer et al., 2018). Single nights at three amounts also shortened latency to persistent sleep in 20 adults with major depressive disorder and persistent insomnia despite antidepressants (Brooks et al., 2019). Models: randomized human studies.

In a larger trial, 364 adults with insomnia disorder received one of three nightly amounts, placebo or zolpidem for 14 days. Night 1 latency to persistent sleep, the primary outcome, and wake after sleep onset in the first 6 hours both showed significant trends with amount (adjusted P < .001). The two higher amounts shortened latency compared with placebo, and on night 13 their effects held while zolpidem's diminished.

Adverse events occurred in 33.8% on seltorexant, 49.3% on placebo and 42.5% on zolpidem; four seltorexant participants discontinued after asymptomatic electrocardiogram-related events (Mesens et al., 2025). A 2022 network meta-analysis found seltorexant more efficacious than placebo in acute use and well tolerated, but its data too scarce for firm conclusions (De Crescenzo et al., 2022). Models: a randomized trial and a meta-analysis.

Seltorexant depression trials

In a randomized, double-blind trial in 47 adults with major depressive disorder, post hoc analysis showed 10 days of seltorexant, but not diphenhydramine, significantly reduced core Hamilton item scores versus placebo, an effect maintained to 28 days without overt changes in the hypnogram (Recourt et al., 2019). A double-blind, adaptive phase 2b trial added placebo or one of three daily amounts to SSRIs or SNRIs in adults with an inadequate response to them.

The middle amount lowered Montgomery-Åsberg (MADRS) scores by 4.5 points more than placebo at week 3 (P = .003) and by 3.1 points at week 6, the primary time point (P = .083), with a larger difference when baseline insomnia was more severe (Savitz et al., 2021). Models: randomized human trials.

As monotherapy, in a phase 1b trial analyzing 86 placebo lead-in non-responders among 128 enrolled, Hamilton score changes at week 5 differed across arms (p = 0.0456), driven by the lower of two amounts versus placebo (p = 0.0049), which the authors call an apparently curvilinear amount-response pattern (Mesens et al., 2025). In an exploratory 24-week phase 2 add-on trial against quetiapine extended release, time to all-cause discontinuation, the primary endpoint, did not differ significantly (41.2% versus 47.1% discontinued), with adverse events in 65.4% versus 80.8% (Pinter et al., 2026). Models: randomized human trials.

Development status

A 2026 drug profile described mixed phase 3 findings, modest effect sizes and gaps on anhedonia, suicidality and long-term safety, pending confirmatory data (Di Nicola et al., 2026). Sources: reviews.