GLP-1 + Naltrexone for AUD: The Stacking Evidence

Two of the better-studied pharmacotherapies for alcohol use disorder — naltrexone (Anton 2006 COMBINE^[5], Garbutt 2005 Vivitrol^[6]) and, much more recently, semaglutide (Hendershot 2025 JAMA Psychiatry^[1]) — act on different pathways and have no pharmacokinetic interaction. That is why obesity-medicine and addiction-medicine clinicians are increasingly stacking a GLP-1 with either oral naltrexone 50 mg or extended-release Vivitrol 380 mg IM for patients who carry both diagnoses. This article walks through the trial data on each component, the real-world EHR signal for semaglutide and AUD (Wang 2024 Nature Communications^[3], Qeadan 2025 Addiction^[4]), and the practical stacking protocol providers are using today.

The honest summary

• Semaglutide reduced heavy-drinking days in a randomized trial. Hendershot 2025 (JAMA Psychiatry^[1]) randomized 48 adults with AUD to semaglutide 0.5 / 1.0 mg weekly or placebo for 9 weeks; the semaglutide arms had significantly lower weekly alcohol craving, lower drinks per drinking day, and a roughly 40% reduction in heavy-drinking days vs placebo.
• Naltrexone is FDA-approved for AUD with decades of data. The COMBINE study (Anton 2006 JAMA^[5]) and the Vivitrol pivotal trial (Garbutt 2005 JAMA^[6]) established naltrexone as a first-line option; Jonas 2014 (JAMA^[7]) meta-analyzed 122 studies and reported a number-needed-to-treat of about 20 for return-to-any-drinking and 12 for return-to-heavy- drinking on oral naltrexone.
• The mechanisms are complementary, not redundant. Naltrexone is a mu-opioid receptor antagonist that blocks ethanol-induced endorphin reward. Semaglutide modulates central reward and satiety via the NTS and VTA dopamine pathway. Stacking targets two independent reward circuits.
• Real-world data support the GLP-1 signal. Wang 2024 (Nature Communications^[3]) analyzed electronic health records of nearly 84,000 patients with obesity and a prior AUD diagnosis and reported a lower incidence and recurrence of AUD diagnoses among those prescribed semaglutide compared with non-GLP-1 anti-obesity medications. Qeadan 2025 (Addiction^[4]) replicated the pattern across substance-related outcomes in a larger GIP/GLP-1 RA cohort.

What Hendershot 2025 actually showed

Hendershot 2025 (JAMA Psychiatry^[1]) is the first randomized placebo-controlled trial of a GLP-1 for alcohol use disorder in humans. The trial enrolled 48 adults meeting DSM-5 criteria for AUD who were not seeking treatment, and randomized them to once-weekly semaglutide titrated to 0.5 mg or 1.0 mg, or matching placebo, for 9 weeks. Primary outcomes were lab-measured alcohol consumption during a self-administration paradigm, plus naturalistic drinking by daily-diary.

Key findings:

• Heavy-drinking days fell about 40% on semaglutide versus placebo over the trial period — the headline outcome that the FDA-relevant pivotal trials for existing AUD drugs use as a key endpoint.
• Drinks per drinking day decreased on the semaglutide arms, consistent with reduced alcohol craving rated on a validated scale.
• Lab-measured self-administered alcohol fell in the cue-reactivity paradigm, suggesting a mechanistic effect on appetitive drive rather than a behavioral substitution.
• A reduction in cigarette smoking was an incidental finding in the smoking subgroup, paralleling the nicotine-related signals seen in EHR cohorts.

The trial is small (n=48) and short (9 weeks), and the dose used (0.5 / 1.0 mg) is below the obesity-treatment dose of 2.4 mg. It is hypothesis-generating, not registrational. But the direction of effect, the dose-response by craving score, and the alignment with the prior exenatide trial (Klausen 2022 JCI Insight^[2]) and the real-world EHR signal (Wang 2024^[3]) collectively make the GLP-1 AUD story one of the most credible new pharmacotherapy signals in addiction medicine in a decade.

The naltrexone evidence: COMBINE, Vivitrol, and Jonas 2014

Oral naltrexone 50 mg was approved by the FDA for alcohol dependence in 1994 based on two earlier trials, and the evidence base was substantially solidified by COMBINE (Anton 2006 JAMA^[5]). COMBINE randomized 1,383 adults with alcohol dependence to naltrexone 100 mg, acamprosate 3 g/day, both, or placebo, each with or without a behavioral intervention, for 16 weeks. Naltrexone with medical management produced significantly fewer heavy-drinking days than placebo; acamprosate did not separate from placebo in this US trial, although it had separated in earlier European trials.

Garbutt 2005 (JAMA^[6]) was the pivotal Vivitrex / Vivitrol trial: 624 actively-drinking adults with alcohol dependence randomized to monthly intramuscular naltrexone 380 mg, 190 mg, or placebo for 24 weeks. The 380 mg arm produced a 25% reduction in heavy-drinking days versus placebo and was the dose later approved as Vivitrol.

Jonas 2014 (JAMA^[7]) systematically reviewed 122 randomized trials and 1 cohort study of pharmacotherapy for AUD in outpatient settings. The meta-analysis reported a number-needed-to-treat of about 12 for return-to-any-drinking and 20 for return-to-heavy-drinking on oral naltrexone, with acamprosate showing a similar effect for return-to-any- drinking. Neither agent is a cure; both reliably reduce drinking days by a meaningful margin.

Mechanism: why the two stack cleanly

Naltrexone is a competitive antagonist at the mu-opioid receptor. Ethanol triggers endogenous opioid release in the ventral tegmental area and nucleus accumbens; mu-opioid antagonism blunts the rewarding subjective effect of a drink, which empirically reduces both craving and drinks-per-occasion.

Semaglutide is a GLP-1 receptor agonist. GLP-1 receptors are densely expressed in the hypothalamic arcuate nucleus, the nucleus tractus solitarius (NTS), and the mesolimbic reward system. Preclinical work in rodents demonstrated that GLP-1 receptor agonism in the VTA reduces alcohol self-administration independently of caloric intake; human imaging extends this by showing reduced cue-induced striatal activation on a GLP-1. The pathways are anatomically and pharmacologically distinct from the opioid system, which is why concurrent blockade of both pathways is mechanistically coherent.

There is no documented pharmacokinetic interaction between naltrexone and semaglutide or tirzepatide. Both are renally cleared substrates with negligible CYP metabolism for the GLP-1 (peptide hydrolysis) and largely glucuronidation for naltrexone.

Magnitude: heavy-drinking-day reduction by intervention

The practical stacking protocol

1. Confirm 7+ days of opioid abstinence before starting naltrexone. Naltrexone is a mu-opioid antagonist and will precipitate withdrawal in opioid-dependent patients. The FDA label requires a documented opioid-free interval (7 days for short-acting opioids, 10–14 days for long-acting). This is the single most common preventable adverse event in naltrexone induction.
2. Pick the naltrexone route by adherence profile. Oral naltrexone 50 mg daily is the starting point for adherent patients; extended-release IM Vivitrol 380 mg every 4 weeks is preferred for patients with a history of medication non-adherence, since one injection covers the next month regardless of daily decisions.
3. Start the GLP-1 once naltrexone is tolerated. Standard semaglutide titration: 0.25 mg weekly × 4 weeks, then 0.5 mg, 1.0 mg, 1.7 mg, 2.4 mg at 4-week intervals. Tirzepatide is reasonable in patients with comorbid type 2 diabetes; the AUD-specific RCT data is semaglutide-only as of 2026, so semaglutide is the off-label-but-evidence-supported choice for AUD.
4. Track AUDIT-C every 4 weeks during titration, then quarterly. The AUDIT-C is a 3-item validated screener; falling AUDIT-C scores parallel the daily-diary drinks-per-day reductions seen in trials.
5. Monitor LFTs at baseline, 12 weeks, and every 6 months. Naltrexone carries a boxed warning for hepatotoxicity at supratherapeutic doses (300 mg/day); the 50 mg AUD dose has a low but non-zero hepatic signal, and any patient with active liver disease deserves quarterly monitoring. GLP-1s have no documented hepatotoxicity and tend to reduce hepatic steatosis alongside weight loss.
6. Document a relapse plan. Neither agent is curative; both reduce the probability and severity of drinking days. Patients should have an explicit plan for a heavy-drinking lapse that includes contacting the prescriber rather than self-discontinuing.

Contrave is not a naltrexone-for-AUD substitute

A common point of confusion: Contrave is a fixed combination of bupropion SR 90 mg and naltrexone SR 8 mg, taken twice daily. The maximum daily naltrexone dose in Contrave is 32 mg, well below the 50 mg AUD-effective dose established by COMBINE (Anton 2006^[5]) and Jonas 2014^[7]. Contrave is approved for weight management and is a reasonable choice for that indication, but it is not a substitute for naltrexone 50 mg in patients with AUD. Patients with both indications either need separate naltrexone 50 mg dosing alongside a GLP-1 (most common in practice) or Vivitrol 380 mg IM every 4 weeks for AUD plus a GLP-1 for weight management.

Cost and access

Generic oral naltrexone 50 mg runs roughly $30/month at most US pharmacies and is on most Medicaid and commercial formularies. Vivitrol (extended-release naltrexone 380 mg IM) lists at roughly $1,500/month but is widely covered by Medicaid AUD benefits and many commercial plans, including as a buy-and-bill office injection. Semaglutide and tirzepatide pricing is unchanged from the obesity-indication list price; AUD is off-label, so insurance coverage typically requires the obesity or T2D primary indication.

AUD-specific telehealth providers (Bicycle Health, Workit Health, Confidant Health, Ria Health) can prescribe oral naltrexone or coordinate Vivitrol injections. GLP-1 prescriptions for the same patient typically come from a separate obesity-medicine prescriber or primary-care clinician, and the two prescribers should communicate the AUDIT-C trajectory and LFT results.

Related research and tools

• GLP-1 and alcohol use disorder — the broader evidence base, the EHR cohorts, and the mechanism in depth
• Naltrexone and Contrave for weight loss — why the 32 mg Contrave naltrexone dose is not an AUD-effective dose
• GLP-1 plus bupropion stacking — the other half of the Contrave story, separated out
• Can you drink alcohol on a GLP-1? — the safety question for patients without an AUD diagnosis
• GLP-1 brain fog and cognitive effects — central-nervous-system effects beyond the reward system

Important disclaimer. This article is educational and does not constitute medical advice. Alcohol use disorder is a serious medical condition; pharmacotherapy should be coordinated with a qualified clinician. Semaglutide is not FDA-approved for AUD as of 2026 and any use for that indication is off-label. Naltrexone induction in opioid-dependent patients requires a documented opioid-free interval to avoid precipitated withdrawal. Patients with active hepatic disease, decompensated cirrhosis, or significant renal impairment require individualized evaluation before either agent is started. PMIDs were verified live against the PubMed E-utilities API on 2026-05-29.

Last verified: 2026-05-29. Next review: every 12 months, or sooner if a registrational GLP-1 AUD trial publishes a primary endpoint.