Oral Semaglutide Cut Heavy Drinking Days in a Phase 2 Trial. It Missed Its Primary Endpoint.
A new randomized trial found fewer heavy drinking days with oral semaglutide, but no significant benefit on its primary craving endpoint. The mixed result is the story.
A pill already prescribed for diabetes may also change how some people drink. In a new randomized trial, oral semaglutide reduced heavy drinking days in adults seeking treatment for alcohol use disorder. That is the encouraging part. The same study missed its primary endpoint, a laboratory test of cue-triggered craving, and it did not significantly reduce drinks per day compared with placebo.
That split result matters more than a simple claim that semaglutide curbs alcohol. It suggests the drug may affect particular parts of drinking behavior without producing a broad, uniform response. It also gives researchers a harder question to answer: are GLP-1 drugs treating alcohol use disorder, changing appetite and reward, or merely making some drinking episodes less intense?
What the oral semaglutide trial tested
The phase 2 trial, published online in The American Journal of Psychiatry on July 29, 2026, enrolled 50 treatment-seeking adults with moderate to severe alcohol use disorder. Participants were randomly assigned to oral semaglutide or placebo for eight weeks. The semaglutide group took 3 mg daily for four weeks, followed by 7 mg daily for another four weeks.
Researchers chose laboratory cue-elicited craving at week six as the primary outcome. In practical terms, they tested whether alcohol-related cues provoked less craving after treatment. Heavy drinking days and drinks per day during the final four weeks were preregistered secondary outcomes. The team also explored craving reported in everyday life, drinks consumed on drinking days, alcohol-related consequences, World Health Organization risk drinking levels, and cannabis use.
Semaglutide did not significantly beat placebo on the primary laboratory craving measure. It also did not significantly reduce drinks per day. Those two negative findings should stay attached to every summary of the trial because a positive secondary outcome cannot retroactively turn a missed primary endpoint into a win.
The signal showed up in heavy drinking
Heavy drinking days did fall significantly with oral semaglutide. The reported model estimate was b=-0.580, with a 95% confidence interval from -1.012 to -0.148. Because the confidence interval did not cross zero, the result met the study's threshold for statistical significance.
Several other measures also favored semaglutide. Drinks per drinking day declined with an estimate of b=-1.177 (95% CI -2.307 to -0.047). Naturalistic craving, meaning craving captured outside the laboratory cue task, fell by b=-2.195 (95% CI -4.174 to -0.216). Alcohol-related negative consequences decreased at a faster rate than with placebo, and more participants moved down by at least one WHO risk drinking level. Cannabis use days also declined in an exploratory analysis.
This pattern is intriguing because it is uneven. A person might still drink on roughly the same number of occasions or report a similar average across all days, yet have fewer heavy episodes and consume less when drinking. That would still be clinically meaningful if a larger trial confirms it. For now, the study tells us where a signal may exist, not how reliably it will translate into treatment.
Why the primary endpoint and daily-life craving disagreed
The laboratory craving task and reports from daily life did not produce the same answer. That is not automatically a contradiction. A controlled cue session captures a narrow response at a specific time. Everyday craving unfolds around meals, stress, sleep, social settings, access to alcohol, and learned habits. Semaglutide could have a detectable effect in that wider setting while leaving a brief laboratory response unchanged.
There are less flattering explanations too. With only 50 participants, the trial may have been too small to estimate several outcomes precisely. Multiple secondary and exploratory analyses also raise the chance that some positive findings will shrink in later studies. The correct reading is neither that the primary endpoint invalidates everything else nor that the positive secondary outcomes prove efficacy. The results are a reason to run a larger, longer study with a clear clinical endpoint.
How this fits with two earlier randomized trials
The oral trial did not appear in a vacuum. A 2025 JAMA Psychiatry phase 2 study randomized 48 adults with alcohol use disorder who were not seeking treatment. Participants received low-dose weekly semaglutide injections or placebo for nine weeks. Semaglutide reduced alcohol consumed during a laboratory self-administration task and lowered peak breath alcohol concentration. It also improved weekly craving, drinks per drinking day, and heavy drinking over time. It did not reduce average drinks per calendar day or the number of drinking days.
A larger 2026 Lancet trial moved closer to ordinary treatment. It enrolled 108 treatment-seeking adults who had alcohol use disorder and obesity. Everyone received cognitive behavioral therapy, while the medication group received weekly subcutaneous semaglutide up to 2.4 mg for 26 weeks. Heavy drinking days fell by 41.1 percentage points from baseline with semaglutide and by 26.4 points with placebo. The estimated treatment difference was -13.7 percentage points (95% CI -22.0 to -5.4; p=0.0015). Transient mild to moderate gastrointestinal adverse events occurred more often with semaglutide.
Taken together, the three randomized trials point in a similar direction on heavy drinking and drinking intensity. They do not show that every alcohol measure improves. They also tested different formulations, doses, populations, treatment settings, and durations. One studied people with obesity receiving therapy, another enrolled people who were not seeking AUD treatment, and the newest tested a lower-dose oral regimen in treatment seekers. That variation makes the repeated heavy-drinking signal more interesting, but it prevents a clean estimate of how much benefit a typical patient should expect.
The broader evidence is still mixed
A June 2026 systematic review found only five eligible clinical studies, although their combined sample reached 49,892 because one was a large cohort. The reviewers described modest improvements in selected alcohol outcomes with semaglutide and dulaglutide, while exenatide was negative in the overall alcohol use disorder population and showed signals only in subgroups. They judged the evidence limited and heterogeneous.
Observational data add scale but not certainty. A January 2026 meta-analysis of five cohort studies found GLP-1 receptor agonist use was associated with a 28% lower risk of an alcohol use disorder diagnosis (hazard ratio 0.72, 95% CI 0.59 to 0.89). The estimate for alcohol-related hospitalization was not statistically significant (HR 0.76, 95% CI 0.57 to 1.01). People prescribed GLP-1 drugs differ from untreated people in many ways, so those associations cannot establish that the medication caused the outcome.
What semaglutide might be changing
GLP-1 signaling is best known for glucose regulation, slowed gastric emptying, and appetite. Addiction researchers are interested in its effects on reward-related behavior. The human trials do not yet identify one proven mechanism for reduced drinking. Lower appetite, altered reward valuation, nausea, changes in alcohol's subjective effects, or a combination could all influence consumption. A treatment can improve behavior before researchers fully map the mechanism, but mechanism matters when deciding who may benefit and which adverse effects are acceptable.
The oral result adds a useful clue. The doses were 3 mg and 7 mg daily, which differ from the higher oral dose used for weight management and from weekly injections used in the earlier alcohol trials. A signal at these doses may suggest that maximal weight loss is not required for an alcohol effect. That remains a hypothesis, not a conclusion, because this study was not designed to separate weight change, gastrointestinal effects, and direct reward effects.
What patients and clinicians should not conclude
Semaglutide is not established as an alcohol use disorder medication on the basis of these trials. The newest study lasted eight weeks and included 50 people. It missed its primary endpoint. Its abstract does not provide the detailed adverse-event table needed for a full safety assessment. Longer studies must test durability, relapse, adherence, psychiatric outcomes, and whether benefits persist when the drug stops.
The evidence also does not support buying unapproved semaglutide or adjusting a prescription without medical supervision. Alcohol use disorder can involve dangerous withdrawal, liver disease, mood symptoms, medication interactions, and urgent safety risks. People taking semaglutide for diabetes or obesity should discuss changes in alcohol use and side effects with the clinician managing that prescription. People seeking AUD treatment deserve a plan built around established addiction care rather than a self-directed experiment.
The Oria take
This is one of the more persuasive GLP-1 addiction signals so far because it comes from a randomized trial in people who wanted treatment. It is also a useful lesson in reading medical news. The trial was positive for heavy drinking days and several related measures, but negative for its primary endpoint and for drinks per day. Both halves belong in the headline.
The next decisive study should be larger, run longer, and choose a clinically direct primary outcome such as heavy drinking days, abstinent days, or relapse. It should report weight change and gastrointestinal symptoms alongside alcohol outcomes, test whether benefit differs by obesity status, and compare medication plus standard therapy with standard therapy alone. Until then, oral semaglutide belongs in the research pipeline, not in a do-it-yourself alcohol protocol.
Sources
Evidence grade: B-. Three randomized trials now show signals on heavy drinking or drinking intensity, including two studies in treatment seekers. Confidence remains limited because the trials are small, differ in formulation and population, and the newest oral trial missed its primary endpoint.
Medical disclaimer: This article is for education only and is not medical advice. Semaglutide is not established or approved as a treatment for alcohol use disorder based on the evidence discussed here. Do not start, stop, or change a prescription without a qualified clinician. Alcohol withdrawal can be dangerous and may require urgent medical care. Oria exists to help readers understand emerging evidence, not to replace diagnosis, addiction treatment, or individualized medical care.
Related Compounds
Build Your Personalised Protocol
Turn your research into a personalised supplementation protocol with the BioStack Generator.