Stress Makes You Want a Drink. But Not Everyone's Brain Works the Same Way, and the Treatment System Doesn't Know That Yet.

On May 2, 2026, The Lancet published results from a randomized, double-blind, placebo-controlled trial of once-weekly semaglutide in adults with alcohol use disorder and comorbid obesity. The trial found that semaglutide significantly reduced heavy drinking days, drinks per drinking day, and self-reported cravings versus placebo. The drug is not FDA-approved for alcohol use disorder — semaglutide is approved for type 2 diabetes and weight management — but the Lancet trial is the most rigorous test yet of GLP-1 agonists for AUD, and its results are compelling enough that the NIH and Veterans Affairs launched a dedicated Phase III trial in May 2026 enrolling veterans with alcohol use disorder.

The semaglutide story is part of a broader shift in how researchers understand alcohol craving, and a 2026 study published in Alcohol, Clinical and Experimental Research adds a layer the prescribing system isn’t ready to act on: stress-induced craving doesn’t work the same way in everyone. The study randomized young men to acute social stress conditions and found that cravings split into two distinct patterns — some participants showed significant craving increases under stress, others showed decreases — and the two groups differed systematically on measures of anxiety, depression, and emotion dysregulation. It isn’t just that some people drink to cope with stress. It’s that the brain mechanism driving the craving differs by psychological profile, and the right treatment differs accordingly.

This matters because the two major pharmacological treatments for AUD target different mechanisms. Naltrexone, which blocks mu-opioid receptors, reduces the reward effect of drinking — it works best for people whose craving is driven by anticipated pleasure (positive reinforcement drinking). Acamprosate, which modulates glutamate activity and reduces withdrawal-associated discomfort, is theorized to work better for stress-relief drinkers: people who are drinking not to feel good but to feel less bad. Prescribing naltrexone to a stress-predominant drinker, or acamprosate to a reward-predominant drinker, isn’t necessarily wrong — but it misses the mechanism. A system that treats AUD as one disease with one neurobiological driver is going to achieve roughly the outcomes a one-size-fits-all approach would predict.

Yale researchers published findings this year showing that higher drinking frequency directly corresponds to lower synaptic density in people with AUD — among the first studies to demonstrate alcohol’s damage to human synapses using PET imaging. The damage accumulates. The longer the gap between onset of problematic drinking and effective treatment, the more structural the change.

That gap is long. Fewer than 2 percent of people with AUD receive any FDA-approved medication, according to NIAAA data. A Frontiers in Psychiatry study found that only 23.3 percent of prescribers reported prescribing naltrexone in the previous three months. Naltrexone has been FDA-approved for AUD since 1994. The barrier isn’t evidence. It’s a prescribing culture that routes people with alcohol use disorder to residential programs and twelve-step groups rather than to physicians who match medication to mechanism.

Semaglutide’s GLP-1 mechanism may work partly by reducing the reward salience of alcohol at the neural circuit level — the same mechanism that appears to reduce food cravings. If so, it might work across both positive-reinforcement and stress-relief drinking patterns, which would explain results that look better than naltrexone alone in trials where the patient population hasn’t been stratified by craving type. Whether that hypothesis holds in the VA Phase III trial will take two years to find out.

In the meantime: the science is ahead of the treatment system by a decade, and the treatment system’s failure to catch up is costing people synapses.

---

Alcohol use disorder research is tracked at /newsroom/substances/alcohol.