Semaglutide and Cannabis Use Disorder: The Wang Study That Changed the Conversation

Cannabis is the most widely used illicit drug in the United States. More than 45 million Americans use it, and roughly one in three of those users meets the criteria for cannabis use disorder (CUD). Despite that prevalence, there are zero FDA-approved medications for treating CUD — a pharmacological desert.

In March 2024, a research team at Case Western Reserve University published a study in Molecular Psychiatry that cracked the door open. Their finding: semaglutide was associated with significantly lower rates of both new CUD diagnoses and CUD relapse in real-world populations. Here's what the data actually shows.

The Study at a Glance

William Wang, Nora Volkow (director of the National Institute on Drug Abuse), and colleagues analyzed electronic health records from the TriNetX Analytics Network — a federated research database spanning approximately 105.3 million patients across 61 large healthcare organizations in the U.S.

They ran two parallel analyses: one in patients with obesity, one in patients with type 2 diabetes. Each cohort was propensity-score matched to control for demographics, comorbidities, and other medications. Follow-up was 12 months.

The Data in Detail

Obesity Cohort

The primary analysis compared 85,223 patients with obesity: those prescribed semaglutide versus those prescribed non-GLP-1RA anti-obesity medications. The mean age was 51.3 years and 65.6% were women. After propensity-score matching ensured balanced groups, the results were striking:

Type 2 Diabetes Cohort (Replication)

To test consistency, the team replicated the analysis in a much larger cohort: 596,045 patients with type 2 diabetes, comparing semaglutide against non-GLP-1RA anti-diabetes medications.

The incident CUD finding in the diabetes cohort was even stronger — a 60% lower risk. The recurrent CUD finding trended in the same direction but didn't quite reach statistical significance (upper CI crossed 1.0), likely due to smaller sample size in the relapse subgroup.

Consistency Across Subgroups

One of the most compelling aspects of this study is how consistent the signal was across every subgroup the researchers examined. The reduced risk held across men and women, across all age groups, across racial categories, and in patients both with and without type 2 diabetes. When a signal is that uniform, it's harder to attribute to confounding.

Why This Matters for Cannabis Specifically

The Wang/Volkow paper is careful to note that their findings are observational and don't establish causation. But the biological plausibility is strong. GLP-1 receptors are expressed throughout the brain's mesolimbic reward pathway — the same circuitry that cannabis activates via the endocannabinoid system. While no preclinical studies have specifically tested GLP-1 agonists against THC (the primary psychoactive compound in cannabis), the cross-substance pattern of reduced addictive behavior in preclinical models — including nicotine, alcohol, cocaine, and opioids — suggests the mechanism is broad rather than substance-specific.

The Broader Pattern

This study doesn't exist in isolation. The same research team (Wang, Volkow, Xu, and colleagues at CWRU) has published parallel analyses showing semaglutide associated with reduced risk of alcohol use disorder, opioid overdose, and tobacco use disorder. Each study uses the same rigorous methodology — large EHR datasets, propensity-score matching, multiple comparator medications, and replication across disease cohorts.

Taken together, the pattern is consistent: semaglutide appears to reduce addictive behavior across multiple substance categories, not just one. This aligns with the hypothesis that GLP-1 receptor agonists modulate reward processing at a fundamental level rather than acting on substance-specific pathways.

Limitations to Keep in Mind

Additional caveats worth noting: CUD diagnosis in EHR data depends on clinicians coding it, which may undercount actual cases. The study cannot measure cannabis consumption directly — only whether a formal diagnosis was recorded. And while propensity-score matching balances known confounders, it cannot account for factors not captured in the medical record.

What's Next

A Phase 2 randomized crossover trial testing tirzepatide (a dual GLP-1/GIP agonist) for cannabis use disorder has been registered at Brigham and Women's Hospital (NCT07265752). This will be the first interventional trial specifically targeting CUD with a GLP-1 class medication.

Until RCT data arrives, the Wang study remains the strongest signal that GLP-1 medications may offer something new for the 16 million Americans with cannabis use disorder — a population that currently has no pharmacological treatment options at all.