Glucagon-like Peptide-1 Receptor Agonists: A New Frontier in Treating Alcohol Use Disorder

The therapeutic potential of GLP-1RAs in AUD was first suggested in a cross-sectional study investigating liraglutide in patients with type 2 diabetes, where reduced alcohol intake was observed as a secondary finding [20]. In 2022, Klausen et al. published the first randomized clinical trial (RCT) to investigate the effects of GLP-1RAs on alcohol consumption, brain function, and alcohol craving in patients with AUD [4] (Table 3). Specifically, Klausen et al. evaluated the effects of exenatide in treatment-seeking individuals with AUD. Although exenatide did not significantly reduce heavy drinking days compared to placebo overall, a subgroup with BMI >30 kg/m² experienced notable reductions in both heavy drinking days and total alcohol intake over the past 30 days [4]. In addition, the neuroimaging study revealed that the exenatide group exhibited reduced alcohol cue reactivity in reward- and addiction-related brain regions, especially the ventral and dorsal striatum. Whole-brain fMRI analyses further demonstrated decreased activation in the left caudate, septal area, and right frontal cortex after 26 weeks of exenatide treatment. These findings suggest that exenatide may decrease the brain's reward response to alcohol-related cues. Moreover, single-photon emission computed tomography (SPECT) scans showed significantly lower dopamine transporter (DAT) availability in reward-processing areas among those receiving exenatide compared to the placebo. However, no significant differences were observed in subjective alcohol craving or cognitive performance between groups [4].

Subsequently, Probst et al. conducted a randomized controlled trial to investigate the effects of dulaglutide on alcohol consumption during smoking cessation [21] (Table 3). The study reported a 29% reduction in alcohol use compared to the placebo [21]. Importantly, these benefits occurred regardless of the smoking status, indicating that GLP-1RAs may have independent effects on alcohol use. It should be noted that this trial was initially designed to study the effect of dulaglutide on smoking cessation. Thus, the study participants did not, per se, suffer from AUD, and the subgroup of heavy drinkers was too small to provide conclusive evidence [21].

More recently, Hendershot et al. published a phase 2 trial in 2025, evaluating semaglutide in non-treatment-seeking patients [22]. This study found that low-dose semaglutide significantly reduced alcohol cravings, the number of drinks per drinking day, and heavy drinking episodes, with medium to large effect sizes. However, it did not impact the number of drinking days or average drinks per day. These mixed but encouraging results highlight the complexity of treating AUD and the need to tailor interventions to individual patterns of alcohol use and motivation for treatment [22]. Notably, participants in both Probst and Hendershot's studies were not seeking treatment for AUD [21,22], underscoring the potential utility of GLP-1RAs in broader, real-world clinical contexts. The vast majority of individuals (~90%) with AUD do not seek formal treatment [23]. The efficacy of these medications in non-treatment-seeking populations suggests that they may still provide benefits when prescribed for other conditions [24].

Beyond clinical trials, observational studies using electronic health records and national registries have provided real-world evidence supporting the therapeutic potential of GLP-1RAs for AUD. Wium-Andersen et al. examined whether the use of GLP-1RA was associated with a decreased risk of alcohol-related events by utilizing data from nationwide registers in the Danish population [25]. Alcohol-related events were defined as (1) hospital contacts with a main diagnosis of AUD (international classification of diseases [ICD]-10 code DF10) in the Danish National Patient Registry, (2) registered treatments for alcoholism in the National Registry of Alcohol Treatment, or (3) the purchase of the benzodiazepine chlordiazepoxide (ATC code N05BA02), which is used for alcohol withdrawal syndrome or the purchase of a medication against alcohol dependence (ATC code N07BB), registered in the Danish National Prescription Registry. GLP-1 receptor agonist users (n = 38,544) and dipeptidyl peptidase-4 (DDP-4) inhibitor users (n = 49,222) were included in the analysis. This study indicates that, after controlling for covariates, using GLP-1 receptor agonists was linked to a lower risk of a subsequent alcohol-related event than using DPP-4 inhibitors. It should be highlighted that starting GLP-1 receptor agonist therapy was linked to a decreased risk of an alcohol-related event when compared to the time without treatment; however, this association was only seen in the first three months of treatment. Nonetheless, this study concluded that GLP-1RAs did not appear to be viable alternatives to existing treatments for AUD (Table 3) [25].

In the United States, Qeadan et al. analyzed data from over 817,000 individuals with a documented history of AUD from the Oracle Cerner Real-World Data. This study discovered that prescriptions for gastric inhibitory polypeptide (GIP) receptor agonists (tirzepatide; dual agonist for GLP1 and GIP) and GLP-1RAs (albiglutide, dulaglutide, exenatide, liraglutide, lixisenatide, semaglutide) were linked to a 50% decrease in alcohol intoxication events compared to those without a prescription for GIP/GLP-1 RA. Specifically, subjects with AUD who had a prescription for GIP/GLP-1 RA had an incident alcohol intoxication rate that was 49%, 42%, and 42% lower, respectively, when stratified by type 2 diabetes, obesity, and type 2 diabetes and obesity (Table 3) [26].

Wang et al. conducted a large retrospective cohort study using electronic health records from the TriNetX Platform. Of the 83,825 obese patients without a prior diagnosis of AUD, 45,797 received their first prescription for semaglutide and 38,028 received non-GLP-1RA anti-obesity drugs, such as topiramate or naltrexone, between June 2021 and December 2022. Following propensity score matching, the two cohorts were balanced (n = 26,566 in each group, mean age 51.2 years, 65.9% women, 15.8% black, 66.6% white, 6.5% Hispanic). Matched cohorts were followed for one year after the index event. In contrast to non-GLP-1RA anti-obesity drugs, such as topiramate or naltrexone, semaglutide significantly reduced the risk of incident and AUD relapse in patients with obesity or type 2 diabetes (Table 3) [8].

More recently, Lähteenvuo et al. analyzed Swedish nationwide registries to determine whether GLP-1RA use decreased the risk of hospitalization for AUD [4]. Among 227,868 individuals with AUD, semaglutide and liraglutide were associated with the lowest risks of hospitalization for both AUD and other SUDs. In contrast, other GLP-1RAs did not show similar effects. Surprisingly, except for naltrexone, FDA-approved AUD medications (acamprosate and disulfiram) were not associated with reduced hospitalization risk. Additionally, these medications were linked to a small but significant increase in suicide attempt risk (adjusted hazard ratio (aHR), 1.15, 95% CI, 1.08–1.22), whereas GLP-1RAs, including semaglutide, were not (semaglutide: aHR, 0.55, 95% CI, 0.23–1.30) (Table 3) [27]. These findings should be interpreted cautiously due to differing comparator groups and potential confounding [27].

While these observational studies cannot establish causality, their consistency across diverse populations and study designs strengthens the case for continued investigation of GLP-1RAs as therapeutic agents for AUD. GLP-1RAs are noteworthy as important candidates for upcoming randomized trials in AUD, as the epidemiological evidence in the metabolic and psychiatric domains has come together.

The most reported side effects are gastrointestinal in nature—primarily nausea, vomiting, and diarrhea. These symptoms are generally dose-dependent and temporary but can lead to dehydration and, in rare instances, pre-renal acute kidney injury [28]. Despite the therapeutic potential of GLP-1RAs, concerns have been raised regarding potential neuropsychiatric side effects, including depressive symptoms and suicidality. Klausen et al. reported a case who died from suicide two months after discontinuing semaglutide; however, no causal link was addressed [4]. Similarly, several observational studies have also raised concerns about a potential link between GLP-1RAs and the increased risk of depressive symptoms and suicidality [29,30]. However, more recent evidence counters these concerns. Specifically, a large nationwide cohort study conducted in France found no short-term increase in suicide risk among users of GLP-1RAs [31]. Similarly, a meta-analysis of 27 randomized controlled trials reported no significant association between GLP-1RA use and suicidality [32]. Additionally, Lähteenvuo et al. reported that semaglutide was not associated with increased suicide risk, whereas FDA-approved AUD medications were linked to a modestly increased risk [27]. Additional studies have, likewise, found no clear association between GLP-1RA use and suicidality [33,34,35,36,37], further supporting the overall neuropsychiatric safety of this drug class.

Clinical trials investigating the role of GLP-1RAs in AUD are still in the early stages, but initial findings are encouraging. GLP-1RAs are thought to influence motivation and reward-related behaviors primarily by modulating the brain's reward system, particularly dopamine signaling pathways in the mesolimbic circuit [38,39,40]. GLP-1 receptors are expressed in key regions involved in reward processing, including the nucleus accumbens and ventral tegmental area. The activation of these receptors has been shown to reduce drug-seeking behaviors [41,42]. Human neuroimaging studies, such as those by Klausen et al., confirmed these findings by demonstrating decreased activation in alcohol cue-reactive brain regions following exenatide administration [4]. Despite no significant reduction in heavy drinking days in patients with AUD overall, when patients were stratified by obesity status, as determined by BMI > 30, significant reductions in heavy drinking days and monthly alcohol intake were observed in patients with BMI > 30. Although the mechanisms underlying this differential response remain unclear, prior research has shown that pharmacokinetic and pharmacodynamic responses to GLP-1RAs vary between lean and obese individuals [43]. Specifically, the administration of intravenous exenatide resulted in an 18.5-fold increase in insulin secretion in lean individuals, as opposed to an 8.8-fold increase in obese individuals [43]. Exenatide, however, dramatically reduced the fMRI signal in the frontal cortex, insula, hippocampus, and amygdala in obese people in response to food pictures but had no effect on the brain fMRI signal in lean participants [43], indicating that the metabolic phenotype may modulate responses of the central nervous system.

Observational and real-world studies using large datasets and national registries provide compelling evidence of the association between GLP-1RA use and reduced alcohol-related outcomes. Studies from Denmark, the United States, and Sweden consistently found lower rates of alcohol-related hospital visits, intoxication events, and AUD recurrence among individuals taking GLP-1RAs compared to those on other anti-diabetic or anti-obesity medications (Table 3). The convergence of findings across diverse populations and methodologies strengthens the case for a real therapeutic signal and suggests that these medications may exert beneficial effects across a range of real-world clinical contexts, including patients with comorbid conditions, i.e., obesity and type 2 diabetes. While observational studies are valuable for detecting real-world patterns, they cannot establish causality and may be subject to confounding.

GLP-1RAs are generally well-tolerated, with gastrointestinal symptoms being the most common adverse events [42]. However, concerns have been raised about potential neuropsychiatric side effects, particularly suicidality. Although isolated case reports and some observational studies suggest a possible link [42,44], the majority of evidence, including large cohort studies and meta-analyses, has not demonstrated a significant increase in suicide risk [29,30]. In contrast, Wang et al. reported that semaglutide may reduce the risk of recurrence suicidal ideation in patients with overweight or obesity when compared to patients treated with non-GLP-1RA anti-obesity medication (HR: 0.44, 96% CI: 0.32–0.60, n = 865 each group after propensity score matching) [8]. Some data suggest that GLP-1RAs may pose a lower suicide risk compared to traditional AUD medications, a finding that warrants further investigation [34]. In a retrospective study conducted using the US Department of Veterans Affairs databases (n = 1,955,135), Xie et al. found that the use of GLP-1RA was associated with a reduced risk of a series of SUDs, including AUD, opioid use disorder, cannabis use disorder, and stimulant use disorder [45]. Additionally, the risk of suicidal ideation, attempt, or intentional self-harm was lower in patients with diabetes who were taking GLP-1RA compared to those who were taking other anti-diabetes medications, such as sulfonylureas, DPP-4 inhibitors, or sodium–glucose cotransporter-2 inhibitors [45]. Moreover, a recent meta-analysis study showed that GLP-1RAs induced significant reductions in the depression rating scales compared to control treatments [46]. These studies added to the body of evidence on the potential benefits of GLP-1RAs in the treatment of neuropsychiatric disorders. They also stress the need for more investigation into the biology and effectiveness of GLP-1RAs as a primary or adjuvant treatment for the treatment of SUDs, psychotic disorders, and depressive disorders.

To date, three double-blind, randomized, placebo-controlled clinical trials have investigated the use of GLP-1RAs in AUD (Table 3). Notably, these studies have small sample sizes and short follow-up durations. Furthermore, the heterogeneity in alcohol use patterns, treatment-seeking behavior, and comorbidities among patients complicate efforts to draw broad conclusions about efficacy. To evaluate the long-term effects of GLP-1RAs on depressive symptoms, the study durations of the included studies listed in Table 3 might not be adequate. The main diagnoses (obesity, type 2 diabetes, or AUD), the GLP-1RA agents and dosages, and the control treatments varied amongst the included studies. Despite these limitations, the available evidence suggests that GLP-1RAs hold therapeutic promise for individuals with AUD.

A major limitation in current GLP-1RA clinical research is the frequent exclusion of individuals with comorbid psychiatric conditions, including co-occurring substance use disorders, depression, or suicidal ideation. Many trials enforce strict exclusion criteria for safety reasons, commonly omitting participants with recent psychiatric hospitalization, activity, a history of suicidal ideation, or a history of suicide attempts. While these precautions are ethically sound, they reduce the generalizability of the findings and may obscure the true risk profile of GLP-1RAs in real-world AUD populations, where psychiatric comorbidities are prevalent. For example, individuals with AUD face elevated risks of both depression and suicidality, yet those most at risk are often underrepresented or entirely excluded from trials [2,47,48]. This is evident in the study by Hendershot et al [22]. (see Table 3), which excluded participants with recent suicidal ideation, a history of suicide attempts, or psychiatric hospitalization within the previous six months. Although large-scale observational studies and meta-analyses to date have not found a significant association between GLP-1RA use and suicidality, existing trials may lack sufficient power or population diversity to detect such risks in individuals with AUD. Therefore, more inclusive and longer-term studies are needed to better assess both the therapeutic potential and safety of GLP-1RAs in this high-risk population.

The efficacy of GLP-1RAs in treating AUD and metabolic diseases varies across agents [27,49], suggesting a need for individualized treatment strategies. Precision psychiatry is an emerging field that aims to tailor mental health care to individuals. To date, only three medications (disulfiram, acamprosate, and naltrexone) have been approved by the United States FDA for the treatment of AUD in the United States. More than 30 different drugs have been tested in alcohol clinical trials in the last 30 years [50]. Most of those clinical trials, though, either did not show any effect at all or showed effects that were very small [50]. This may be due to the heterogeneity of AUD phenotypes and the critical knowledge gaps that underlie the pathophysiology of AUD and the mechanisms of action of the medications used to treat AUD [51,52]. Although DSM-5, a symptom-based tool, is used to assist in the diagnosis of mental disorders, its accuracy is not questionable. However, psychiatry lacks biological tools to evaluate or predict clinical outcomes using biological and objective measures [53,54]. The promise of precision psychiatry lies in understanding the complex interplay of biological and environmental factors, ultimately leading to personalized diagnosis, treatment, and prevention strategies [55,56].

GLP-1RAs are believed to influence alcohol-related behaviors by modulating the brain's reward system, particularly dopaminergic signaling in the mesolimbic pathway [57]. To further elucidate the underlying mechanisms responsible for individual variation in drug treatment response, the patient-derived induced pluripotent stem cell (iPSC) model system offers a valuable platform for investigating the cellular and molecular effects of GLP-1RAs in the central nervous system [58,59]. Importantly, these personalized models can recapitulate human brain tissue, which could be a useful tool to determine individual variability in drug response, potentially guiding future drug development and clinical trial design.

The effectiveness of GLP-1RAs varies, and it has been noted that this is due, at least in part, to differences in the molecular formula (Table 2). Consequently, the half-life and PK/PD profiles are different, suggesting a need for individualized treatment strategies [27,49]. Each drug, even within the same drug class, may possess distinct molecular profiles and mechanisms of action, thereby highlighting the necessity of patient-derived in vitro cell models to evaluate drug action in the brain [60,61,62]. Although current in vitro assays and in vivo models designed to discover potential therapeutic targets in psychiatry are useful, there is an urgent need to establish patient-derived model systems to complement studies that use other models for neuropharmacology research [63,64,65]. The patient-derived iPSC model system offers distinct strengths, including (1) working with cells that retain the patients' unique genetic background, (2) the ability to recapitulate human brain tissue, and (3) the ability to experimentally manipulate live brain-like cells, which is the beauty of the iPSC-based cell model system. Understanding of how GLP-1RAs work at the cellular and molecular levels could potentially identify new drug targets and drug repurposing opportunities for AUD. The patient-derived iPSC cell model could catalyze a paradigm shift that will add more mechanistic rigor to future clinical trials and revolutionize clinical practice for the treatment of AUD. Furthermore, it is also critical to identify central and peripheral biomarkers (e.g., neuroimaging, blood-based markers) that can predict or monitor individual responses to GLP-1RA treatment in AUD.

With the technology evolving, it is conceivable to uncover novel therapeutic targets for AUD by employing a multi-omics framework [66,67] and advanced systems biology techniques and using patient-derived iPSC model systems and drugs as molecular probes, i.e., GLP-1RAs [3,60,61,62]. These research tools could offer a robust and reliable capability to detect potential pharmacological targets for future investigations.

One of the major challenges that we are currently dealing with is not a lack of data but rather how we use the data that is already available from biobanks, public databases, clinical trials, preclinical studies, and real-world observation studies to generate a testable hypothesis. We should also point out that novel biological system network algorithms, machine learning, and artificial intelligence platforms have shed new light on disease mechanisms and underlying drug mechanisms, addressing critical challenges in big data-oriented biomedically complex systems [68,69,70,71]. These tools have already led to novel discoveries, thus laying the groundwork for testing novel therapeutic agents [62,72,73]. In addition to recommending larger, longer-term clinical trials, we also emphasize the importance of precision psychiatry approaches—such as the identification of predictive biomarkers, the development of patient-derived iPSC models, and the application of AI and multi-omics tools—to address the heterogeneity of AUD and variability in treatment response to GLP-1RAs. These advanced technologies can enhance mechanistic understanding, enable individualized treatment strategies, and guide more targeted clinical trial designs.

Altogether, the current literature indicates that GLP-1RAs represent a promising new direction in the pharmacologic management of AUD. Their central action on reward pathways, combined with robust real-world evidence of reduced alcohol-related harm and a generally favorable safety profile, suggests that GLP-1RAs could have clinical utility beyond metabolic diseases. As these findings continue to evolve, future research should prioritize larger, longer-term RCTs that evaluate GLP-1RAs across diverse patient populations, with careful attention to treatment motivation, psychiatric comorbidities, and long-term outcomes.