Almost weekly there is a new claim. These medications prevent Alzheimer’s. They cure alcoholism. They rebuild cartilage. Microdosing them extends life.
Some of these claims sit on phase 3 trials with tens of thousands of participants and regulatory approvals. Some sit on a single small study in mice. Some are influencer extrapolation built on a kernel of real mechanism.
This article walks the major non-weight-loss claims being made about GLP-1 medications, in rough order from best evidenced to most speculative, and grades how strong the actual science is. The goal is not to recommend a treatment. The goal is to map the evidence honestly, with the strongest source for each claim noted.
This is a living document. The science here is moving faster than almost any area of medicine, claims that look speculative today may be settled in a year, and claims that look solid may not survive the next large trial. So we treat this piece as a standing state of play rather than a one-time article. We review and republish it roughly monthly, adding new claims as they surface, regrading the existing ones as evidence arrives, and flagging the confident-but-wrong claims that tend to circulate online in between. If you are reading a version of this, it reflects the best evidence we had at the date at the top. Check back; it will have changed.
First, what these drugs actually are
GLP-1, glucagon-like peptide 1, is a hormone the gut releases when you eat. It does several things at once. It tells the pancreas to release insulin. It tells the pancreas to suppress glucagon, the hormone that raises blood sugar. It slows how fast the stomach empties. And it acts on receptors in the brain to signal satiety.
GLP-1 receptor agonists are synthetic molecules that mimic this hormone but last far longer in the body. Natural GLP-1 has a half-life of around two minutes; the modern weekly medications last about a week. The class has a few broad types:
- Single agonists that act on the GLP-1 receptor alone.
- Dual agonists that act on both the GLP-1 receptor and the GIP receptor, a related gut hormone.
- Older daily-injection agonists, the first generation of the class.
- Triple agonists, acting on GLP-1, GIP and glucagon receptors, currently in late-stage trials.
These are not weight-loss molecules. They are receptor agonists that happen to produce weight loss as one downstream effect. GLP-1 receptors are expressed throughout the body, in the pancreas, brain (including reward circuits), kidneys, heart, immune cells, and vascular endothelium. Activate those receptors systemically with a long-acting drug and you get a cascade of effects, of which appetite suppression is only the most visible.
That framing matters. It explains why the same molecule keeps showing up in trials for conditions that look unrelated. The receptor is everywhere. The drug acts on all of it.
How to read the evidence grades
For each claim below, the evidence is graded:
- Strong: multiple large randomised controlled trials, consistent effect sizes, often with regulatory approval for the indication.
- Moderate: at least one well-designed RCT showing benefit, with supporting observational and mechanistic data.
- Emerging: small RCTs, large observational studies, or strong preclinical signals not yet confirmed in humans.
- Speculative: mechanism plausible, but evidence is preclinical, anecdotal, or observational with significant confounding.
- Negative: directly tested and failed.
1. Cardiovascular disease (strong)
Likely mechanism: plausible contributors include anti-inflammatory, vascular, metabolic, and haemodynamic effects, which is why some researchers argue this class is better understood as a cardiometabolic disease-modifying therapy than simply a weight-loss drug.
The SELECT trial randomised 17,604 adults with established cardiovascular disease and a BMI of 27 or higher, but no diabetes, to a single-agonist GLP-1 medication or placebo. The result was a 20 percent reduction in major adverse cardiovascular events, a composite of cardiovascular death, non-fatal heart attack, and non-fatal stroke.
The interesting finding came in the follow-up analyses. The cardiovascular benefit did not track with weight loss. People with early dramatic weight loss did no better than those with modest weight loss. Mediation analysis suggested roughly a third of the cardiovascular benefit was explained by waist circumference reduction; the remaining two-thirds came from somewhere else. Benefits held across pre-specified subgroups including patients with prior heart failure and prior coronary artery bypass grafts.
Bottom line: for people with established cardiovascular disease and excess weight, GLP-1 medications reduce future heart attacks and strokes. This is the indication with the strongest evidence outside of diabetes itself.
2. Chronic kidney disease (strong)
Likely mechanism: reduced hyperfiltration, improved glomerular haemodynamics, and anti-inflammatory effects in the kidney.
The FLOW trial enrolled 3,533 adults with type 2 diabetes and chronic kidney disease. The GLP-1 medication reduced the primary composite kidney outcome, kidney failure, a 50 percent or greater decline in kidney function, or kidney or cardiovascular death, by 24 percent. It also reduced major cardiovascular events by 18 percent and all-cause death by 20 percent. The effect held across the spectrum of CKD severity, including advanced CKD where most diabetes drugs become difficult to use. Regulators approved the indication on the strength of these results in early 2025.
Bottom line: for type 2 diabetics with kidney disease, this class now sits alongside ACE inhibitors, SGLT2 inhibitors, and finerenone as foundational therapy. Evidence outside diabetes is thinner but consistent.
3. Metabolic dysfunction-associated steatohepatitis (MASH, strong)
Likely mechanism: reduced liver fat, reduced systemic inflammatory drivers, and emerging evidence of direct action on liver sinusoidal endothelial cells.
MASH is the inflammatory, scarring stage of fatty liver disease. It is the leading cause of liver transplant waiting lists in developed countries, and until recently there was essentially no pharmacological treatment.
The ESSENCE trial randomised 1,197 adults with biopsy-confirmed MASH and moderate-to-advanced fibrosis. At 72 weeks, 62.9 percent of those on the GLP-1 medication achieved MASH resolution without worsening fibrosis, compared to 34.3 percent on placebo. Around 37 percent achieved at least one stage of fibrosis improvement. When researchers controlled for weight loss, the medication still showed liver-specific improvements, suggesting direct hepatic effects. Since the liver itself has very few GLP-1 receptors, the leading hypothesis is action on liver sinusoidal endothelial cells and reduced systemic inflammation. Regulators granted accelerated approval for MASH with fibrosis in 2025, and a dual agonist is showing similar or better results in the SYNERGY-NASH trial.
Bottom line: the first effective drug class for one of the most common liver diseases in the world. The long-term outcomes data, preventing cirrhosis, liver cancer, transplant, will take another five years to mature.
4. Obstructive sleep apnoea (strong)
Likely mechanism: reduced fat deposits around the upper airway and tongue, plus weight-independent effects on hypoxic burden and inflammation.
The SURMOUNT-OSA trial randomised adults with moderate-to-severe obstructive sleep apnoea and obesity to a dual agonist or placebo for 52 weeks. The apnoea-hypopnoea index, the standard measure of how often you stop breathing per hour of sleep, fell substantially. Up to 51.5 percent of participants met the criteria for disease resolution. The medication also reduced the hypoxic burden, a stronger predictor of cardiovascular mortality than the apnoea count itself. Regulators approved the indication in late 2024, the first drug ever approved specifically for sleep apnoea.
Bottom line: for the substantial overlap population (around 60 to 70 percent of OSA patients have obesity), this is now a real treatment option, particularly for people who cannot tolerate CPAP.
5. Knee osteoarthritis (moderate, with an interesting twist)
Likely mechanism: mostly mechanical (less load on the joints), but emerging evidence of direct metabolic effects on cartilage cells.
The STEP 9 trial randomised 407 adults with obesity and moderate-to-severe knee osteoarthritis. Pain (measured by the WOMAC score) dropped 41.7 points on the GLP-1 medication versus 27.5 on placebo. Function improved. NSAID and opioid use went down.
The obvious explanation was less weight on the joints. But that never quite fit the clinical pattern, patients reported pain relief well before significant weight loss. A small follow-up published in Cell Metabolism in February 2026 suggests something more interesting: in both mice and a small human pilot, the medication appeared to shift cartilage cell metabolism from inefficient glycolysis toward oxidative phosphorylation, reduce cartilage degeneration, and improve joint space, even when weight was held constant. If this holds up in larger trials, the class could be classified as disease-modifying for osteoarthritis, not just symptomatic. The cartilage-restoration data is still preliminary. The pain and function data is solid.
Bottom line: strong evidence for symptom relief in obese OA patients. Genuinely promising but unconfirmed evidence for structural disease modification.
6. Alcohol use disorder (moderate)
Likely mechanism: modulation of mesolimbic reward circuitry, the same brain pathway that drives food, alcohol, and drug reward.
GLP-1 receptors are expressed in the brain’s reward centres, and animal studies have consistently shown these agonists reduce alcohol self-administration. A 2025 randomised trial in JAMA Psychiatry (Hendershot et al.) enrolled 48 adults with alcohol use disorder for nine weeks. The medication reduced alcohol consumption in a laboratory self-administration test, reduced weekly craving, and, in a subgroup of smokers, reduced cigarettes per day. A small phase 2 trial, but the signal was clear enough to justify the phase 3 trials now underway. A separate 26-week Copenhagen trial in adults with AUD and obesity showed reduced heavy-drinking days. Large real-world data (Wang et al., Nature Communications 2024) found a 50 to 56 percent lower incidence of new AUD diagnoses among users compared to matched controls on other anti-obesity medications.
Bottom line: evidence is consistent across preclinical, observational, and small-RCT levels, all pointing the same direction. Of the speculative claims, this is the area where the gap between press confidence and trial confidence is smallest.
7. Other addictions (emerging)
Likely mechanism: the same reward-circuit modulation as alcohol, applied to nicotine, opioids, gambling, and compulsive eating.
Observational data and case reports suggest reductions in nicotine use, opioid cravings, gambling, and compulsive shopping. The Hendershot trial picked up a cigarette signal as a secondary finding. There are ongoing trials in cocaine and opioid use disorder. Outside alcohol and nicotine, the human RCT data is sparse.
Bottom line: plausible mechanism, scattered signal, no definitive trials yet.
8. The “acetaldehyde trap” on triple agonists (speculative, and probably wrong)
The claim, as it circulates: that the newest triple agonists (acting on GLP-1, GIP and glucagon receptors) create a dangerous “acetaldehyde trap” when you drink, by suppressing the liver enzyme that clears alcohol, so toxic acetaldehyde piles up and the hangover or harm is amplified. Some of the figures attached to this online are specific, a 40 to 60 percent reduction in the relevant enzyme.
This is worth walking through carefully, because it’s a good example of the kind of confident, mechanistic-sounding claim that turns out to be muddled on inspection, and because the triple agonists are new enough that very little is actually settled.
Start with what is reasonably solid. Like the rest of the class, a triple agonist slows gastric emptying. That has a real consequence for alcohol: it hits the stomach and bloodstream on a slower, less predictable timeline, and it irritates the gastric lining for longer. So drinking on one of these medications can mean worse and more prolonged nausea, and a higher, more erratic blood-alcohol curve. That interaction is mechanistically sound and consistent with how the whole class behaves. It’s the part worth respecting.
The acetaldehyde-specific claim is where the reasoning falls apart. The enzyme being pointed to, CYP2E1, is only a minor pathway for turning alcohol into acetaldehyde. The main route is alcohol dehydrogenase, and the actual bottleneck that causes acetaldehyde to accumulate sits downstream, at a different enzyme (ALDH2). So even if a triple agonist did suppress CYP2E1, the effect would, if anything, be slightly less acetaldehyde produced, not more. The “perfect storm of accumulation” framing has the biochemistry backwards. The specific 40 to 60 percent figure appears to be borrowed from general GLP-1 literature rather than drawn from any triple-agonist trial data. It reads as marketing-grade reasoning from peptide-vendor sources, not an established finding.
And the better evidence points the opposite way on alcohol generally. In animal work, both triple and dual agonists have attenuated alcohol-seeking behaviour, which is precisely why these molecules are being studied as potential treatments to reduce drinking, the same thread as the alcohol use disorder section above, not as agents that make alcohol more toxic.
Bottom line: there is no good evidence that a triple agonist meaningfully increases acetaldehyde or creates a special toxic interaction with alcohol. The real, well-grounded interaction is gastrointestinal, slower emptying plus shared dehydration making for nastier, longer nausea and a more erratic blood-alcohol curve. That is the thing to respect. The acetaldehyde trap is a phantom. We flag it here precisely because it is the kind of claim our evidence grading exists to catch, and because, as the triple agonists move through trials, this is a section we expect to revisit as real data arrives.
9. Cancer prevention (emerging)
Likely mechanism: likely indirect, reduced hyperinsulinaemia (which is mitogenic), reduced systemic inflammation, reduced obesity itself (a known cancer risk factor), and possibly altered gut microbiota.
A large 2026 retrospective study presented at the ASCO Gastrointestinal Cancers Symposium analysed 281,656 matched patients and found GLP-1 users had a 36 percent lower incidence of colorectal cancer than aspirin users. In high-risk subgroups the reduction was around 42 percent. A separate ASCO 2025 study suggested reductions across 14 obesity-related cancers in diabetic patients. These are observational; propensity matching helps but does not eliminate confounding, and cancer endpoints have a way of disappearing when you move from observational to randomised data. Earlier safety signals had pointed the other way, pancreatic and thyroid cancer concerns from the original labels, and while most have not been borne out in long-term data, they have not entirely vanished either.
Bottom line: intriguing observational signals, especially for colorectal cancer. Not yet confirmed in randomised trials. Treat as hypothesis-generating.
10. Systemic inflammation (moderate on mechanism, emerging on clinical translation)
Likely mechanism: direct action on GLP-1 receptors expressed in immune cells, plus indirect effects via reduced adipose-tissue cytokine release.
The most consistent finding across the literature, after the metabolic effects, is reduction in inflammatory markers. C-reactive protein drops reliably across trials in obesity and type 2 diabetes; a 2025 meta-analysis of 40 RCTs in 6,029 patients found CRP reduction was the most consistent inflammatory effect. In large trials, only about 20 to 60 percent of the CRP reduction could be explained by weight loss and glucose control, implying a substantial direct anti-inflammatory effect. This is one of the leading hypotheses for why cardiovascular benefits appear independent of weight loss, and may explain emerging signals in conditions like psoriasis.
Bottom line: the anti-inflammatory effect is real and biologically substantial. Whether it translates into clinical benefit for specific inflammatory conditions is mostly still being tested.
11. Fertility, especially PCOS (emerging)
Likely mechanism: improved insulin sensitivity restoring ovulation in metabolically driven infertility, not a direct hormonal effect.
Polycystic ovary syndrome is the most common cause of ovulatory infertility, and it is fundamentally a metabolic condition driven by insulin resistance. Several small trials and meta-analyses suggest GLP-1 agonists improve menstrual regularity, ovulation rates, and natural pregnancy rates in women with PCOS. The phenomenon of unexpected pregnancies on these medications appears to be real, driven by restored ovulation and possibly by reduced absorption of oral contraceptives due to delayed gastric emptying. Early sperm-parameter studies in men also show improvements.
Important caveat: these drugs should not be used during pregnancy. Most clinicians recommend stopping well before conception. Early-pregnancy exposure data from a large cohort is reassuring, but the long-term safety database is still small.
Bottom line: likely to help fertility in metabolically driven infertility, particularly PCOS. Stop before conception. Contraception planning matters more than people realise.
12. Mood and anxiety (mixed signal, recently clarified)
Likely mechanism: initial reports were likely a statistical artefact of post-marketing surveillance bias, not a drug effect.
In 2023 to 2024, post-marketing reports raised concerns that these medications might increase suicidal ideation. Regulators investigated. In April 2026, regulators requested removal of the suicidal-behaviour warning from GLP-1 receptor agonist labels, based on a comprehensive meta-analysis of 91 placebo-controlled trials covering 107,910 patients, which found no increased risk versus placebo. Separate large meta-analyses, multiple Mendelian randomisation studies, and large observational cohorts have all converged on the same answer: no increased risk of suicide or self-harm. Some propensity-matched observational data even suggests an association with reductions in major depression, though that is observational and likely confounded.
Bottom line: the initial safety signal did not survive larger analysis. As of mid-2026, the regulatory consensus is no increased psychiatric risk. This is a sensitive area; anyone experiencing low mood or thoughts of self-harm should speak with their clinician or a support service rather than navigating it alone.
13. Alzheimer’s disease (mostly negative, one bright spot)
Proposed mechanism: GLP-1 receptors in the brain, reduced neuroinflammation, improved insulin signalling. The hypothesis was plausible. The trial data has not co-operated.
The EVOKE and EVOKE+ trials, the largest GLP-1 trials ever conducted in Alzheimer’s disease, randomised 3,808 adults with early-stage Alzheimer’s to an oral GLP-1 medication or placebo for two years. Results presented at the December 2025 CTAD conference were disappointing: the medication did not slow disease progression versus placebo. Some biomarkers improved, but the clinical endpoint failed. A smaller phase 2b trial of an older daily agonist (the ELAD trial, Nature Medicine December 2025) showed the opposite pattern in 204 patients: nearly 50 percent less brain volume loss on MRI and an 18 percent slower decline in cognitive function over a year. The catch is that ELAD’s primary endpoint, cerebral glucose metabolism, was not met; the MRI and cognitive findings were secondary and exploratory, and reviewers have flagged possible selection bias.
Bottom line: on current evidence, this class does not slow established Alzheimer’s disease. The older-agonist signal keeps the door slightly ajar for prevention or earlier-stage use, but it is one trial against a much larger failed one.
14. Longevity and all-cause mortality (speculative in healthy people)
Proposed mechanism: if the cardiovascular, metabolic, and anti-inflammatory benefits seen in high-risk populations extend to healthy ones, lifespan should follow. That extrapolation has not been tested.
Both SELECT and FLOW showed reductions in all-cause mortality, roughly 19 to 20 percent. These are real, in real populations, but those populations were high-risk: established cardiovascular disease, or diabetes with kidney disease. The speculative leap, increasingly common in biohacker and microdosing circles, is that these medications will extend lifespan in healthy people. There is currently no clinical trial data testing this, and such a trial would need to be very large, very long, and would almost certainly never be commercially funded. The case against extrapolation is that the side-effect profile, lean mass loss, GI effects, possible bone density loss, might matter more in healthy populations where the upside is smaller.
Bottom line: mortality benefits are real in high-risk populations. Longevity claims in healthy people are speculation built on extrapolation. The trials that would settle this are not running.
Two side effects worth getting right
The two most-discussed downsides of GLP-1 medications are lean mass loss and bone density loss. Both are real. Both are also routinely misrepresented. They deserve the same evidence treatment as the benefits above.
Lean mass loss: real, but the framing is mostly wrong
Cause: rapid weight loss of any kind reduces lean mass proportionally. This is biology, not a class-specific effect.
The headline number is true: roughly 25 to 40 percent of the weight lost on this therapy is lean mass. What gets lost in the headlines is what that number actually means. First, the same ratio shows up in essentially every weight-loss intervention ever studied, caloric restriction, bariatric surgery, lifestyle change. A network meta-analysis of 22 RCTs and 2,258 participants found lean mass loss accounted for around 25 percent of total weight loss, the same proportion you see with diet and exercise alone. Second, the relative composition of the body, lean as a percentage of total mass, is essentially unchanged. People do not become disproportionately less muscular. They become proportionally smaller. Third, the fat that comes off is not generic fat. The STEP 1 DEXA substudy found that at 15 percent total body weight loss, fat mass dropped 19.3 percent and visceral fat, the metabolically dangerous deep abdominal kind, dropped 27.4 percent. Visceral fat is what drives cardiovascular and diabetes risk. Losing it preferentially is the entire metabolic story.
So why does the “these drugs waste muscle” framing have such grip? Two reasons. The absolute scale of weight loss is large, so even a normal proportion produces a substantial absolute lean mass loss. And for some populations the absolute loss matters more, older adults already losing muscle to age-related sarcopenia have less margin. The mitigation is the same as for any rapid weight loss: 1.6 to 2.2 g/kg of goal body weight in protein, progressive resistance training two to three times a week, and avoiding faster-than-1-percent-per-week loss where possible.
Bottom line: lean mass loss is real but proportional. The relative body composition is preserved. The visceral fat loss is the upside that makes the absolute lean loss worth tolerating, and the protein-plus-resistance-training fix is straightforward.
Bone density loss: real, modest, and weight-loss-driven
Cause: mechanical unloading of the skeleton during rapid weight loss, not a direct drug effect on bone.
This one is genuinely worth attention if you are in the wrong demographic, and overblown if you are not. The strongest direct evidence comes from Hansen et al. 2024 (eClinicalMedicine), a 52-week double-blind RCT in 64 adults at elevated fracture risk: hip bone mineral density dropped 2.6 percent versus placebo, lumbar spine 2.1 percent. Larger-scale data from the STEP 1 bone substudy (Obesity 2023) showed similar but smaller effects, around 1.7 percent total hip BMD decrease over 68 weeks, proportional to weight loss magnitude.
The cause is mechanical, not pharmacological. Bone responds to load. Less body weight means less load. The skeleton remodels downward to match the new mechanical demands, the same effect seen in astronauts, bedridden patients, and after bariatric surgery. The clinical implications track demographics: for pre-menopausal adults with normal baseline bone density it’s modest and likely clinically irrelevant; for postmenopausal women, older adults, or people with existing osteopenia it’s meaningful and worth managing. A 2024 trial showed combining the therapy with resistance and weight-bearing exercise preserved hip and spine bone density at placebo-equivalent levels.
Bottom line: modest bone density loss driven by mechanical unloading, not the drug itself. Clinically important in older and postmenopausal populations. Manageable with screening, resistance training, and supplementation. Not a reason to avoid these medications, but a reason to plan around them.
Where this leaves us
GLP-1 medications are genuinely one of the most consequential drug classes of the last twenty years. The cardiovascular, kidney, liver, and sleep apnoea evidence is strong enough that, for the populations who qualify, withholding them would now be hard to defend.
The alcohol use disorder data is the area most likely to deliver another major indication in the next two years. The osteoarthritis cartilage data, if it replicates, could reframe joint disease. Alzheimer’s looks unlikely on current evidence. Cancer prevention is a maybe. Longevity in healthy people is a story, not a finding.
The side effects most often raised, lean mass and bone density, are real but mostly proportional to weight loss itself, not specific to the drug class, and largely manageable with resistance training, protein, and where indicated baseline screening.
In short: these are not weight-loss drugs that happen to do other things. They are systemic receptor agonists with effects on metabolism, inflammation, reward circuitry, vascular function, and kidney physiology, and weight loss is one downstream consequence. That framing predicts most of what is showing up in the trials. It also predicts that some of the more speculative claims will turn out to be real, and some will not.
The drugs are powerful. The hype runs ahead of the data, as it always does. The data is still mostly catching up, and it is, in most areas, catching up in the direction of “yes, but in this specific population, for this specific outcome.”
That is more interesting than a wonder drug. It is a real one.