Semaglutide Just Passed Its Hardest Test Yet — And the FDA Has No Roadmap for What Comes Next

Picture a clinical coordinator at a Copenhagen investigational site, twelve weeks into a randomized double-blind trial, watching a participant report their drinking log for the week: zero heavy-drinking days. Not a rounding error. Not a measurement artifact. The participant had entered the trial with an AUDIT score above 15 and at least six heavy-drinking days in the four weeks before screening. The coordinator marks the electronic diary entry and moves to the next participant. This happens again. And again. By week 26, the pattern has hardened into data — and that data now sits in The Lancet, demanding a response from a regulatory system that has no existing channel for it.

The trial enrolled 108 adults between 18 and 70 years old, all carrying a dual burden: alcohol use disorder and comorbid obesity defined by a BMI of 30 kg/m² or higher. The gender split was nearly equal — 51% men. Participants received once-weekly subcutaneous semaglutide or placebo, mirroring the dose escalation protocol established in the obesity indication. The primary and secondary outcomes captured both alcohol consumption metrics and body weight. What the investigators found wasn’t a marginal signal that requires statistical massaging to see. It was robust enough to land in The Lancet on the first submission cycle.

But behind the publication fanfare lies a regulatory vacuum that no one in the GLP-1 space seems eager to discuss publicly.

The Mechanism Nobody Ordered

The GLP-1 receptor agonist class was developed to manage glycemic control. Obesity came later, validated through the STEP trial program — in STEP 1, once-weekly semaglutide 2.4 mg produced a mean body weight reduction of 14.9% from baseline over 68 weeks in adults without type 2 diabetes. Cardiovascular risk reduction followed: on March 8, 2024, the FDA approved Wegovy for a new indication — reducing the risk of cardiovascular death, heart attack, and stroke in adults with established cardiovascular disease and obesity or overweight. Each of these expansions followed a legible regulatory path. A sponsor identified a mechanism, designed a confirmatory RCT, negotiated endpoints with the FDA under a Special Protocol Assessment or equivalent pre-submission agreement, and filed a supplemental NDA. The agency reviewed against a known standard.

Alcohol use disorder offers none of that comfort. The FDA has approved naltrexone for AUD — that approval came in 1994, thirty-two years ago — and acamprosate followed, with a network meta-analysis showing naltrexone produces statistically significant reductions in heavy drinking versus placebo. The existing standard of care is modest, utilization rates are poor, and the patient population is chronically underserved. What the FDA has never done is evaluate a GLP-1 receptor agonist for a psychiatric or substance use indication. The CDER division that would handle such a filing — likely the Division of Psychiatry — has no institutional muscle memory for semaglutide’s pharmacology. The Division of Metabolism and Endocrinology Products, which knows Wegovy inside out, has no jurisdiction over AUD.

That jurisdictional gap alone could add eighteen months to any regulatory strategy a sponsor attempts to build around this data.

The epidemiological rationale for targeting this dual population is not manufactured. A study of treatment-seeking AUD patients found that 17.82% were obese and another 36.63% were overweight — meaning more than half of people presenting for alcohol treatment carry excess weight. The Lancet trial specifically recruited this comorbid population, which means its findings apply directly to the patients most likely to walk into an addiction treatment program. The trial design wasn’t reaching for a niche. It was describing a plurality of the AUD treatment-seeking universe.

When Cohort Data Meets RCT Confirmation

Before this Lancet publication, the strongest human evidence linking semaglutide to reduced alcohol consumption came from a nationwide cohort study in Sweden, published in JAMA Psychiatry in November 2024. The study analyzed medical records of 227,868 individuals diagnosed with AUD in Sweden between 2006 and 2021. Among them, 4,321 used semaglutide and 2,509 used liraglutide. The signal was there — but cohort studies carry confounding that randomized trials don’t, and the FDA does not approve new indications on observational architecture alone, regardless of what its 2023 real-world evidence framework says about trial emulation.

The Lancet trial changes the evidentiary standard. A randomized, double-blind, placebo-controlled design — 108 participants, dual-indication population, pre-specified endpoints — converts a pharmacovigilance signal into a mechanistic demonstration. GLP-1 receptors are expressed in dopaminergic reward pathways. The preclinical literature suggested for years that activating those receptors could attenuate the reinforcing properties of alcohol. The Lancet trial didn’t discover a new mechanism. It confirmed one under conditions rigorous enough to withstand regulatory scrutiny.

Consider what that means for Novo Nordisk’s asset map. Wegovy currently sits in a GLP-1 market valued at approximately USD 62.83 billion globally in 2025. An AUD indication — even a narrow one targeting the obese, treatment-seeking subpopulation — would open a channel into addiction medicine that no metabolic drug has ever entered. The commercial logic is obvious. The regulatory logic requires construction from scratch.

The Approval Architecture Doesn’t Exist Yet

Here is the counterintuitive reality sponsors need to sit with: the strength of this Lancet RCT may actually complicate the regulatory path rather than clear it. When a drug has a robust efficacy signal in a population that straddles two CDER divisions, sponsors face a co-review problem. Any sNDA for semaglutide in AUD would require the FDA to determine which division leads the review — and that determination shapes everything: the endpoint standards, the safety database requirements, the labeling conventions, and the risk evaluation and mitigation strategy framework. A REMS designed for metabolic indications looks nothing like a REMS designed for a psychiatric or substance use disorder indication.

The FDA’s posture here is understandable — but incomplete. The agency has issued guidance on psychiatric drug development and separate guidance on the role of biomarkers in addiction medicine, but nothing that bridges a GLP-1 mechanism to an AUD clinical development program. A sponsor entering a Type B pre-IND meeting today would be negotiating in a framework that does not yet have a template for this therapeutic intersection. The FDA would need to build review standards in real time, and that process transfers timeline risk directly onto the sponsor’s development calendar.

That logic holds — until you look at the naltrexone precedent more carefully. Naltrexone’s 1994 AUD approval was itself a repurposing: the drug had been approved for opioid dependence in 1984. The FDA navigated a cross-indication expansion then, and it can navigate one now. But 1994 operated without the dual-division complexity that a GLP-1 filing would trigger, and without the commercial stakes that now attach to any Novo Nordisk regulatory move in a USD 62.83 billion market under active price scrutiny from Congress.

What a sponsor like Novo Nordisk should do — before filing anything, before a confirmatory Phase 3 program is designed — is request a Type B pre-IND meeting that explicitly asks the FDA to adjudicate divisional jurisdiction. That meeting request should arrive with the Lancet paper attached and a draft development framework that proposes a joint review structure. Waiting until an IND is active to surface the jurisdictional question is how programs lose eighteen months to administrative back-and-forth that serves no one. The pre-IND meeting is not a courtesy — it is a structural necessity for any program operating at the intersection of metabolism and psychiatry.

Sponsors should also build the safety database requirements for both frameworks simultaneously. AUD populations carry hepatic comorbidities that standard obesity safety monitoring doesn’t prioritize. Liver function surveillance cadences, alcohol biomarker integration (phosphatidylethanol over self-report), and the question of whether heavy-drinking relapse events constitute a serious adverse event under GCP — these are not details to resolve post-IND. They belong in the protocol before first patient in.

Back in Copenhagen, that coordinator’s database entry — zero heavy-drinking days, week 12 — is now peer-reviewed, published, and publicly available. The preclinical hypothesis has a randomized trial behind it. The cohort signal from 227,868 Swedish patients has a controlled confirmation. Every piece of the scientific architecture is in place.

The only thing missing is the regulatory architecture to receive it — and Novo Nordisk’s next move will determine whether the FDA builds that architecture proactively or gets forced to improvise it one review cycle at a time.

References

1. The Lancet — “Once-weekly semaglutide versus placebo in patients with alcohol use disorder and comorbid obesity: a randomised, double-blind, placebo-controlled trial”
2. Israel Pharm / JAMA Psychiatry (2024) — “Semaglutide and liraglutide repurposing for AUD: nationwide cohort study, 227,868 individuals”
3. Diabetes on the Net — “STEP 1 trial: semaglutide 2.4 mg, 14.9% mean body weight reduction over 68 weeks”
4. FDA Press Announcement — “FDA Approves First Treatment to Reduce Risk of Serious Heart Problems Specifically in Adults with Obesity or Overweight,” March 8, 2024
5. PMC / Pilot Study — “Prevalence of obesity and overweight in treatment-seeking AUD patients: 17.82% obese, 36.63% overweight”
6. Medical Economics — “FDA approval of naltrexone for AUD, 1994; network meta-analysis efficacy data”
7. Fortune Business Insights — “GLP-1 receptor agonist global market valued at USD 62.83 billion in 2025”
8. AJMC — “GLP-1s reduce heavy drinking days in patients with obesity and alcohol use disorder: Lancet trial design details”