The smoking claim did not begin with a marketing department. It began in 2017, with a mouse experiment showing that nicotine itself switches on GLP-1 neurons in the brainstem, and that those neurons feed a circuit whose job is to make an animal stop. Nine years later the human record contains a consistent result, and it is not the one the claim describes: the drugs change what happens to body weight around a quit attempt, and they have not yet been shown to change whether anyone quits.
That distinction matters commercially, because a subscription sold on the second promise while only the first has evidence behind it is the exact pattern set out in the red-flag article.
The circuit is about avoidance, not reward
Smokers titrate their intake to stay below the dose at which nicotine becomes unpleasant, and the 2017 study asked what enforces that ceiling. Nicotine activated GLP-1 neurons in the nucleus tractus solitarius. Sitagliptin and exenatide both cut nicotine intake in mice, and so did chemogenetic activation of those same neurons, while mice lacking the GLP-1 receptor consumed more nicotine than wild-type animals. Optogenetic work then traced the route: GLP-1 excites medial habenular projections to the interpeduncular nucleus, and activating receptors in that pathway abolished nicotine reward and lowered intake, while knocking them down or blocking them raised it.[1] The authors describe the neurons as satiety sensors for nicotine.
That is a different anatomy from the accumbens work that underpins the alcohol indication. It predicts a drug that makes a cigarette feel like enough sooner, not one that makes cigarettes unwanted — which is roughly what the human trials go on to find.
The animal results are consistent, and specific
In mice, exendin-4 at a dose with no effect of its own attenuated nicotine-induced locomotor stimulation, accumbal dopamine release and the expression of conditioned place preference, and blocked the expression of locomotor sensitization.[2] In rats of both sexes trained to self-administer intravenous nicotine for about 21 days, liraglutide cut self-administration and cue-plus-priming reinstatement, and separately reduced both withdrawal-driven high-fat-diet intake and body weight gain, at a dose that did not produce malaise-like effects.[3]
The specificity is worth noting because it cuts against a general anti-addiction story. Tested against opioids in the same kind of abuse-related paradigms, GLP-1 receptor agonist treatment did not reduce the abuse-related effects of those drugs.[4] Whatever this receptor is doing, it is not switching off wanting in general.
Every trial that prespecified quitting has missed
The largest randomized test is a Swiss single-center trial of 255 adults with at least moderate cigarette dependence who wanted to quit. Everyone received behavioral counseling and varenicline at 2 mg a day; on top of that, participants were assigned to 12 weeks of dulaglutide 1.5 mg weekly or placebo. The primary outcome was biochemically confirmed point prevalence abstinence at week 12. It came in at 63% (80 of 127) on dulaglutide against 65% (83 of 128) on placebo, a difference of −1.9 percentage points. Craving fell during treatment with no difference between the groups. Weight fell 1.0 kg on dulaglutide and rose 1.9 kg on placebo, a baseline-adjusted difference of −2.9 kg (95% CI, −3.59 to −2.3; P < 0.001), and glycated hemoglobin fell by a median 0.25 percentage points more on drug.[5]
The 2026 trial built specifically around semaglutide reached the same place from a different direction. It randomized 24 non-treatment-seeking adults smoking at least five cigarettes a day to nine weeks of subcutaneous semaglutide titrated to 1.0 mg, or placebo. Neither co-primary laboratory endpoint separated: smoking resistance at β = 0.16 (95% CI, −0.07 to 0.40; P = 0.16) and number of cigarettes at β = −0.08 (95% CI, −0.25 to 0.08; P = 0.30). Craving fell and body weight fell (β = −0.04; 95% CI, −0.05 to −0.03; P < 0.001).[6]
The one trial widely quoted as positive is a 2021 pilot in 84 prediabetic or overweight smokers, all of whom got a 21 mg nicotine patch and brief counseling, randomized to exenatide 2 mg weekly or placebo for six weeks. Abstinence was 46.3% against 26.8% — a risk ratio of 1.70, with a 95% credible interval of 0.96 to 3.27. Post-cessation body weight was 5.6 pounds lower on exenatide.[7] An interval running from slightly below no effect to more than triple the odds is the result of a 42-per-arm study, and it is why the word pilot is in the title.
The weight effect is real, and it does not outlive the injection
Following the Swiss trial out to a year is the most instructive thing in this literature. Abstinence fell to 43% against 41% at week 24 and 32% against 32% at week 52 — identical. Weight, the endpoint that had separated cleanly, closed as well: the baseline-adjusted between-group difference was −1.0 kg (97.5% CI, −2.16 to 0.16) at week 24 and −0.35 kg (95% CI, −1.72 to 1.01) at week 52, with both groups up about 3 kg from baseline.[8]
Twelve weeks of a weekly injection bought no measurable weight advantage a year out, which is the same rebound pattern documented after stopping for any reason in the discontinuation article. A predefined secondary analysis of the same trial found the mirror image in blood pressure: dulaglutide lowered weight and pressure early, then pressure rose by 7.5 mmHg by week 52 as the weight came back.[9] The 2026 meta-analysis of randomized trials pooled the cessation evidence at an odds ratio of 1.36 (95% CI, 0.55 to 3.35; P = 0.51), graded very low certainty, against a weighted mean difference in post-cessation weight of −4.14 kg (95% CI, −7.22 to −1.05).[10]
The large numbers are diagnosis codes, and they arrive too fast
The figure that launched most of the coverage is a 2024 target trial emulation across 222,942 new users of diabetes medicines, 5,967 of them starting semaglutide, all with both type 2 diabetes and a recorded tobacco use disorder. Against insulin, the hazard ratio for a medical encounter coded as tobacco use disorder within twelve months was 0.68 (95% CI, 0.63 to 0.74); against other GLP-1 receptor agonists it was 0.88 (95% CI, 0.81 to 0.96). Smoking cessation medication prescriptions and counseling fell too.[11]
Two features of that paper deserve to travel with its numbers. The authors record that for most comparisons the difference emerged within 30 days of the first prescription, and they list among the study’s limitations that it has no data on current smoking behavior at all. The outcome is a billing code for a clinical encounter, not a cigarette. A patient who has just started an injectable and is being seen for that reason is a patient whose visit gets coded around the new drug, and a month is a short window in which to stop smoking.
Human genetics does not fill the gap. The multi-ancestry drug-target Mendelian randomization study that found genetically modeled GLP1R and GIPR agonism associated with less binge drinking reported estimates for tobacco that were consistently null.[12] The receptor variation that tracks one behavior does not track this one.
How much randomized evidence exists, exactly
Less than the coverage implies. The 2026 systematic review screened 751 records, reviewed 26 full texts, and found two randomized controlled trials reporting smoking cessation, together enrolling 337 people.[10] The public trial register carries nine studies of a GLP-1 receptor agonist in smoking cessation; the largest is a 300-participant phase 2 trial of tirzepatide that was still recruiting, with a primary completion date in March 2028, and the largest completed one is the 255-participant Swiss trial above. Nothing at the scale that supports a label has been run, let alone reported.
By contrast, varenicline — the comparator that was given to everybody in the Swiss trial, including the placebo arm — is an approved drug with its own evidence base, and it is what produced the 63% and 65% abstinence rates that the injection failed to improve on. Any claim about this class has to clear that bar rather than a placebo one.
What this means for a cash subscription
No GLP-1 receptor agonist is approved by the FDA for smoking cessation or for nicotine dependence, so a prescription written for that purpose is off-label everywhere in the country. Most product sold through cash telehealth is compounded, which is not FDA-approved and is not reviewed by the FDA for safety, efficacy or quality before it is dispensed.
The defensible version of the claim is narrow and still worth something: people who quit smoking gain weight, that weight gain is one of the reasons they start again, and a GLP-1 receptor agonist demonstrably slows it while it is being taken. The Swiss trial showed the drug holding weight down by 2.9 kg through the quit attempt. The same trial showed the advantage gone a year later, once the injections stopped. Anyone weighing that trade should also weigh the side-effect profile in the side effects article, because in the Swiss trial gastrointestinal symptoms affected 90% of the dulaglutide group. How every figure on this site is established before publication is described in the methodology.