Alkermes said on Monday that its experimental compound ALKS 7290 reduced ADHD symptom scores substantially in adults, and the headline number the company reported, a median 19-point drop on a 54-point scale at the highest dose, is the kind of result that moves a stock. The shares rose about 6.5 percent.
The part worth paying attention to is not the size of the drop. It is that the drug works on a system no approved ADHD treatment has ever touched, and that the trial producing those numbers was built to test safety, not efficacy. Both facts are true at once, and holding them together is the whole job of reading this result correctly.
A wakefulness system in an attention disorder
Orexin is a neuropeptide made in the lateral hypothalamus, the brain region that governs wakefulness. It was identified in the late 1990s, and the discovery that narcolepsy with cataplexy is caused by the loss of the neurons that produce it turned a rare sleep disorder into a well-characterized deficiency state. That is what makes the receptor an attractive target: in narcolepsy the disease is the missing signal, so replacing it is a more direct intervention than treating the symptom from downstream. Orexin receptor agonists are among the most advanced experimental treatments for that condition, and Alkermes has late-stage programs in narcolepsy and idiopathic hypersomnia.
ALKS 7290 targets the orexin 2 receptor. According to the company's release, the rationale for using it in ADHD runs through the same circuits from the other direction: orexin engages pathways implicated in attention, cognition and mood, so a compound that increases signaling there might address symptoms that stimulants approach through monoamine transmission.
That would make it the first mechanistically distinct option in a market that has been unusually static. The stimulants prescribed today descend from compounds characterized in the 1930s and 1950s, and the non-stimulant alternatives, atomoxetine and viloxazine, also act on norepinephrine, a monoamine. A drug that works by promoting wakefulness signaling rather than by blocking reuptake is not a refinement of the existing class. It is a different class, which is why a safety study generated this much attention.
The class already has a marketed drug
Orexin agonism stopped being speculative in August. The FDA approved Orzeyful, Takeda's oveporexton, on August 5 for narcolepsy type 1, making it the first orexin-targeted therapy to reach the market and the first narcolepsy treatment aimed at replacing the missing signal rather than compensating for it downstream. Takeda's own announcement describes a twice-daily tablet and projects $2 billion to $3 billion in annual sales in that indication alone.
A marketed predecessor matters for two reasons. It means the mechanism is no longer a bet that biology is on the right track, because patients are already taking a drug built on it. It also means ADHD is the first real test of whether the class travels beyond sleep. Everything approved or in late-stage development for orexin so far treats disorders of excessive sleepiness, and the field is well funded: Lilly paid $6.3 billion for Centessa, Eisai and others have their own orexin programs, and more than a dozen molecules are in development. Alkermes is using ADHD to argue the system does something broader, and its own pipeline makes the same argument with a second compound in development for fatigue associated with multiple sclerosis and Parkinson's disease.
If the Phase 2 works, the franchise question answers itself. If it does not, the class stays a sleep story.
What fifty patients can and cannot show
The Phase 1 program had three parts. Two tested single and multiple ascending doses in 88 healthy volunteers for up to 10 days and did not find a maximum tolerated dose, which is what the company means when it describes a wide therapeutic index. The third part, the Phase 1b, enrolled 50 adults with ADHD.
The design was careful. Participants stopped their existing ADHD medications for a two-week washout, then were randomized 4:1 within each of two cohorts to receive split daily doses of either 20 milligrams, 50 milligrams, or placebo, for 14 days of inpatient treatment. Enrollment split into 20 participants at each dose and 10 on placebo.
The primary objective was safety and tolerability. The symptom scales were exploratory endpoints, and the company states plainly that the study was neither designed nor powered to detect statistically significant differences between treatment groups. There was no placebo-adjusted comparison reported, because a 10-person placebo arm in an exploratory study cannot support one.
That is not a flaw the company is hiding. It is what a first-in-patient study is for. But it means the result is a signal about a mechanism, not a measurement of an effect, and the distinction is what separates this announcement from a Phase 3 readout.
The improvements are large, which is why the next trial matters more
With that caveat stated, the numbers are not trivial. Participants entered with a median score of about 39 on the Adult ADHD Investigator Symptom Rating Scale, a clinician-administered instrument scored out of 54, which describes a substantially symptomatic group. At day 14, median reductions were 14.0 points at 20 milligrams and 19.0 points at 50 milligrams, with improvements appearing as early as day 6, the first post-baseline assessment.
On the Clinical Global Impression-Severity scale, a seven-point rating of how ill a clinician judges the patient to be, baseline medians of 4.0 and 5.0 in the two dose groups fell by 1.0 and 2.0 points, moving participants from the markedly or moderately ill range into the mildly ill range.
The study also ran EEG-based biomarkers and performance-based cognitive tests, and reported treatment effects across processing speed, information processing, working memory and attention. Those measures matter more than they might appear to. Symptom rating scales depend on a clinician's judgment and a patient's report, and a mechanism that leaves no trace on objective cognitive measures would be harder to believe. A drug that changes both the rating and the EEG is at least doing something measurable to the brain.
Safety looked manageable in the treated groups. No serious treatment-emergent adverse events were reported, most side effects were mild, and the most common were insomnia, frequent urination, dizziness, changes in sustained attention, urinary urgency and constipation. No participant receiving ALKS 7290 discontinued; two on placebo did.
Insomnia and urinary symptoms are worth naming rather than passing over, because they are the predictable signature of a drug that increases wakefulness signaling. The approved narcolepsy label for Orzeyful lists insomnia and increased urinary frequency and urgency among its most common adverse reactions, which is the same cluster ALKS 7290 produced. That cuts both ways. It means the profile is a known property of the class rather than a surprise specific to this molecule, and it also means the class carries a tolerability cost that regulators have already judged worth writing down.
A treatment for an attention disorder that keeps patients up at night is trading one problem for another, and whether the trade is acceptable is exactly what a larger and longer trial has to establish. Fourteen days of inpatient treatment in 40 people on active drug cannot answer it, and neither can a rating scale administered by the same investigators who know which patients are on placebo in a study this small.
Stimulants are the standard, and their supply has been the problem
The context that makes a new mechanism more than an academic curiosity is the state of the existing supply. Amphetamine and methylphenidate products have spent years on the FDA's drug shortage list, and the current snapshot includes extended-release methylphenidate tablets, mixed amphetamine salts and lisdexamfetamine capsules. When the first-line class is intermittently unavailable, patients and clinicians absorb the disruption through dose changes, pharmacy calls and substitutions between formulations that are not always interchangeable.
A second class would not fix a manufacturing problem. Nothing about orexin biology addresses why a generic stimulant goes on backorder. But a non-stimulant with a different mechanism is the kind of option that becomes more valuable when the stimulant supply is unreliable and when patients do not tolerate monoaminergic drugs well. The clinical need Greg Mattingly, a psychiatrist at Washington University School of Medicine, described in the release is for "differentiated therapies," and differentiation by mechanism is the kind that survives a supply shock.
What the Phase 2 has to prove
The confirmatory trial is already running. It will enroll about 312 adults with ADHD, test once-daily and split dosing against placebo after the same two-week washout, and use change from baseline in the AISRS total score at week four as its primary endpoint. The first participant was dosed this month, and the company expects data in 2027.
Three things about that design will decide whether ALKS 7290 is a drug or a hypothesis. Placebo response is large and rising in ADHD trials, which is why the field has seen well-run programs fail late, and a drug judged on attention is measured against a symptom that moves with sleep, stress and the expectations of the person reporting it. Four weeks is short enough that any durable benefit has to emerge quickly, and long enough for a placebo response to build. And a trial powered for a primary endpoint produces an effect size with a confidence interval rather than a pair of median reductions, which is the number that finally settles whether the drug does what the small study suggests it might.
Alkermes has another reason to want the answer quickly. Its commercial products treat alcohol dependence, opioid dependence, schizophrenia, bipolar I disorder and narcolepsy, and its late-stage pipeline sits in narcolepsy and idiopathic hypersomnia. An orexin agonist that works in ADHD would extend the franchise from sleep disorders into psychiatry, which is the strategic logic behind the way the company's chief medical officer, Craig Hopkinson, framed the result: as support for exploring orexin biology in disorders beyond excessive sleepiness.
That is a large claim resting on a 50-patient safety study. The field will know more in 2027, and until then the reading that survives scrutiny is that a new mechanism has passed its first test without yet having been tested for the thing that matters.
Primary sources
- Alkermes for the ALKS 7290 trial results, dose arms, AISRS and CGI-S figures, adverse events, and the statement that the study was not powered to detect statistically significant differences between groups.
- ClinicalTrials.gov for the Phase 1 program structure, enrollment and the ongoing Phase 2 design.
- FDA for the August 2026 approval of Orzeyful (oveporexton) and its indication.
- Takeda for Orzeyful's dosing, mechanism, adverse reactions and sales projections.
- FDA drug shortages database for the current stimulant shortage snapshot.