GLP-1 Drugs: What They Fix and What They Don't

I.1Where it comes from

GLP-1 (glucagon-like peptide-1) is a 30-amino-acid peptide made by L-cells, hormone-releasing cells scattered through the lining of the lower small intestine (the ileum) and the colon. They sense nutrients in the gut: fat, protein, glucose, and the short-chain fatty acids that gut bacteria make from fibre. When food arrives, they release GLP-1 into the tissue fluid around them.¹

The same gene, proglucagon, is cut differently in different tissues. In the pancreas it becomes glucagon. In L-cells it becomes GLP-1 and GLP-2, released together in equal amounts. A small group of neurons in the brainstem (the nucleus of the solitary tract) also makes GLP-1, so the brain has its own supply separate from the gut.¹

I.2Why it only lasts two minutes

GLP-1's second amino acid is alanine, which is exactly what the enzyme DPP-4 (dipeptidyl peptidase-4) recognises. DPP-4 snips off the first two amino acids, and the shortened peptide no longer activates the receptor. The enzyme is everywhere in the blood, but it's concentrated in the lining of the tiny blood vessels that drain the gut. In pigs, GLP-1 stored inside L-cells was 95% intact, while GLP-1 that had just entered the gut's blood vessels was only 33% intact. The authors found DPP-4 in the capillaries sitting right against the L-cells.²

In people, intact GLP-1 injected into a vein or under the skin is cut within minutes. Thirty minutes after an injection under the skin, nearly 90% of what was circulating in diabetic patients was already the inactive fragment.³ The liver removes much of what's left, so only about 10–15% of the GLP-1 your gut releases reaches the general circulation still active.⁴

I.3So it works mostly where it's made

A hormone that's mostly destroyed before it reaches the rest of the body has to act nearby. The gut wall is full of nerve endings from the vagus nerve, which carries signals from the gut to the brainstem, and those endings carry GLP-1 receptors. In rats with those receptors knocked down, meals got bigger, the stomach emptied faster, blood glucose after meals rose higher and insulin release was weaker.⁵ Natural GLP-1 works like a local alarm: food has reached the lower gut, so slow the stomach, release insulin, and stop eating soon.

Two features matter for what follows. First, its insulin effect is glucose-dependent: GLP-1 boosts insulin only when blood glucose is raised, which is why it rarely causes low blood sugar on its own. Second, the slowing of the stomach (the "ileal brake") is a big part of why people feel full.¹

I.4GLP-2, the half the drugs leave out

The L-cell releases GLP-2 alongside every pulse of GLP-1, and GLP-2's job is the gut lining. In mice, ten days of GLP-2 injections tightened the intestinal barrier, cutting the leak of both small molecules between cells and large proteins through them. The functional change appeared within four hours, before any change in the structure of the lining.⁶ GLP-2 also makes the lining grow, and a GLP-2 drug (teduglutide) is used for short bowel syndrome.

GLP-1 drugs activate only the GLP-1 receptor. Someone who eats in a way that triggers their own L-cells gets both signals. Someone on semaglutide gets a large, constant GLP-1 signal and no extra GLP-2. Nothing in the drug repairs a leaky gut.

Strongfor GLP-1's breakdown and half-life (measured directly in people); Animalfor vagal signalling and GLP-2's barrier effect.

II.1From lizard venom to a once-a-week shot

The first GLP-1 drug came from the venom of the Gila monster. In 1992 a researcher isolated exendin-4, a peptide about half identical to human GLP-1 that activates the same receptor. Its second amino acid is glycine instead of alanine, so DPP-4 can't cut it.⁷ Synthetic exendin-4 became exenatide (Byetta, 2005), injected twice a day. Liraglutide (Victoza, Saxenda) came next, with a fatty acid attached, and lasted long enough for one shot a day.

II.2The three changes in semaglutide

Semaglutide (Ozempic, Wegovy, and Rybelsus as a pill) is human GLP-1 with three changes.⁸

1. The enzyme can't cut it. The alanine at position 8 is replaced with Aib (aminoisobutyric acid), an amino acid that isn't in normal proteins. DPP-4 doesn't recognise it.

2. A fatty-acid tail. At position 26, a lysine carries a linker and an 18-carbon fatty diacid. Albumin, the most common protein in blood, normally carries fatty acids, so it grabs this tail. Most of the drug in your blood is riding on albumin at any moment. Bound to something that big, it's too large for the kidney to filter out, and it's shielded from enzymes. It comes off albumin slowly, a little at a time.

3. One more swap at position 34 (lysine to arginine) so the fatty acid only attaches where it's meant to.

The result had a half-life of 46 hours in mini-pigs.⁸ In people it's about a week, which is why it's injected once a week and takes four to five weeks to reach steady levels. Tirzepatide (Mounjaro, Zepbound) uses the same fatty-acid trick on a peptide that activates both the GLP-1 receptor and the receptor for GIP, the other gut hormone that boosts insulin. Its half-life is about five days.

II.3Why lasting longer means reaching more places

Natural GLP-1 comes in short bursts, mostly confined to the gut and the blood leaving it. The drug sits in the blood at high, steady levels all day and night, so it reaches tissues the natural hormone barely touches.

The most important of these is the brain. Neither drug crosses the blood-brain barrier in general. They get in at the circumventricular organs, small areas around the brain's fluid spaces where the barrier is leaky by design so the brain can sample the blood. In mice given labelled liraglutide, the drug showed up in those areas and bound neurons in the arcuate nucleus of the hypothalamus, the main appetite centre. In rats, its weight loss didn't depend on GLP-1 receptors in the vagus nerve, which is the route natural GLP-1 uses.⁹ Semaglutide reached the brainstem, hypothalamus and septal nucleus the same way, changed which foods mice chose, and cut food intake without lowering energy expenditure.¹⁰

The strength-of-signal row is the whole story. The drug isn't "more of your own GLP-1." It's a signal your body never makes, held at a level it never reaches, in places it doesn't normally go. That's why it works so much better than anything that raises natural GLP-1, and it's also why its side effects don't look like anything natural GLP-1 does.

Strongfor the structure and half-life; Animalfor where in the brain it acts (rodents; human brain access is inferred from that and from imaging).

III.1Weight, and where it comes from

In STEP 1 (1,961 adults with obesity, no diabetes), weekly semaglutide 2.4 mg for 68 weeks took off 14.9% of body weight against 2.4% on placebo; half the semaglutide group lost 15% or more.¹¹ In SURMOUNT-1, tirzepatide 15 mg took off 20.9% over 72 weeks, and more than half on the top dose lost a fifth of their body weight.¹² Those numbers approach weight-loss surgery.

The fat that goes is disproportionately the dangerous kind. In the STEP 1 body-scan subgroup, total fat fell about 19% while visceral fat, the fat packed around the organs, fell about 27%.¹¹ In an MRI analysis of a later semaglutide trial, visceral fat fell 1.5 litres more than on placebo.²⁶ Visceral fat drains straight into the liver through the portal vein and is the fat most tied to insulin resistance, fatty liver and heart disease.

III.2Insulin sensitivity

Part of the glucose benefit is GLP-1's direct effect on the pancreas (more insulin, less glucagon, only when glucose is high). Part of it is the body actually responding to insulin better. In a 28-week trial in type 2 diabetes using the clamp method (the gold-standard measure of insulin sensitivity), both tirzepatide and semaglutide improved insulin sensitivity, tirzepatide more so. On tirzepatide, a combined measure of how much insulin the pancreas puts out for a given level of insulin resistance rose about sevenfold.¹³ Losing visceral and liver fat is the likely main route: less fat spilling into the liver means the liver stops pumping out glucose it shouldn't.

III.3Liver

In ESSENCE, 800 people with inflamed fatty liver (MASH, formerly NASH) and moderate or advanced scarring took semaglutide or placebo for 72 weeks. Liver inflammation cleared without worsening scarring in 62.9% versus 34.3%, and scarring improved in 36.8% versus 22.4%.¹⁴ Fatty liver is the condition the Linoleic Acid paper ties most directly to diet.

III.4Heart, and why it isn't just weight

In SELECT, 17,604 people with heart disease and overweight but no diabetes took semaglutide or placebo for an average of about 40 months. Heart attack, stroke or cardiovascular death fell from 8.0% to 6.5%, a 20% relative reduction (HR 0.80).¹⁵ In people with diabetes and kidney disease (FLOW), semaglutide cut major kidney events 24% and cardiovascular death 29%.¹⁷ In heart failure with a stiff heart (STEP-HFpEF), symptoms and walking distance improved and CRP, the main blood marker of inflammation, fell 44%.¹⁸

The follow-up analysis of SELECT is the most interesting part. The heart benefit was the same whatever people weighed at the start, and how much weight people lost in the first 20 weeks didn't predict their later risk. Shrinking waistlines did: about a third of the benefit ran through waist reduction. The other two thirds wasn't explained by fat loss at all.¹⁶ On placebo, people who lost weight actually had more heart events, which is what you'd expect when weight loss comes from getting sick.

So the drug is doing something to blood vessels and inflammation beyond shrinking fat, and the fat it does shrink that matters is the fat around the middle. This is the strongest argument that the drug is good for people, not just their scales.

III.5Cravings beyond food

The appetite centres the drug reaches also handle reward. In a small trial in 48 adults with alcohol use disorder, low-dose semaglutide cut the amount people drank in a lab drinking session, drinks on days they drank and weekly craving. In smokers, it cut cigarettes per day.¹⁹ Many patients describe the same thing as "food noise" going quiet.

Strongfor weight, visceral fat, liver, heart and kidney (large randomised trials); Moderatefor insulin sensitivity by clamp (one 28-week trial) and alcohol (one small trial).

IV.1Fat is a result, not the cause

Nobody becomes obese by accident of willpower alone. Fat accumulates because something upstream is pushing it there: a diet high in linoleic acid that changes how fat cells grow and signal (Linoleic Acid), too little muscle to dispose of glucose, too little movement, poor sleep, chronic inflammation, a damaged gut lining, and food engineered to override fullness. Appetite is high because the body is responding to those inputs.

A GLP-1 drug works at the end of that chain. It turns down appetite, so less food comes in, so stored fat is used. It doesn't lower the linoleic acid in the fat that remains, build muscle, repair the gut lining (I.4), or change what you eat, only how much. Someone who started with low muscle, a leaky gut, chronic inflammation and seed-oil-heavy fat stores, and changes nothing else, ends up a smaller person with low muscle, a leaky gut, chronic inflammation and seed-oil-heavy fat stores.

IV.2Stopping shows what was underneath

If the drug fixed the cause, the weight would stay off when it stopped. It doesn't. In STEP 4, everyone took semaglutide for 20 weeks and lost 10.6%, then half switched to placebo. Over the next 48 weeks the ones who stayed on lost another 7.9%; the ones switched to placebo gained 6.9%.²¹ In the STEP 1 extension, people who had lost 17.3% regained two thirds of it within a year of stopping, and their blood pressure, blood sugar and lipid improvements drifted back toward baseline with it.²⁰

The trial authors read this as proof that obesity is chronic and the drug has to be taken for life. The other reading is that the drug was holding down a drive that nothing else had changed.

IV.3Fasting and dieting have the same problem

Losing weight by not eating, whether by drug, fasting or a hard diet, works on the same end of the chain. The body defends its weight. Fourteen contestants from The Biggest Loser lost an average of 58 kg in 30 weeks of extreme diet and exercise. Six years later they had regained 41 kg, and their resting metabolism was still about 700 calories a day below where it started, about 500 calories lower than their body size predicted.²² The people who kept more weight off had the most metabolic slowing, so the body was still pushing back.

Fasting has its own benefits (see Metabolic Health), but as a way to lose weight it's still a cut in intake. It doesn't change the linoleic acid in your fat, build muscle or fix the gut, and the drive to eat comes back when the fast ends.

IV.4The drug plus a change in inputs

The best trial of what happens when you add something upstream is Danish. After an 8-week low-calorie diet (13.1 kg lost), 195 adults were randomised for a year to placebo, exercise, liraglutide, or liraglutide plus exercise. The combination kept off the most weight and lost twice as much body-fat percentage as either alone. It was also the only group whose insulin sensitivity, long-term blood sugar (HbA1c) and fitness improved.²³ The drug alone kept weight off; only adding exercise changed the metabolism underneath. Gallstones and a raised heart rate were more common on liraglutide alone than in the combination.

Strongfor weight regain after stopping (randomised withdrawal); Strongthat exercise adds metabolic benefit on top of the drug; Inferredthat the drug leaves gut barrier, adipose LA and inflammation drivers untouched (from how it works; not measured as outcomes).

V.1How much of the weight is muscle

Body scans split weight into fat, bone and "lean mass," which includes muscle but also organs, water, glycogen and the water inside fat tissue. In the STEP 1 scan subgroup, lean mass fell about 10% while fat fell 19%, so roughly two fifths of the weight lost was lean.¹¹ In SURMOUNT-1 it was closer to a quarter. Across trials the share ranges from about 15% to as much as 40–60%.²⁴

Two things soften that. First, lean mass isn't the same as muscle. Some of it is liver shrinking as fat leaves it, water lost with glycogen, and the support tissue of fat that's no longer needed. Second, any weight loss costs some muscle; dieting alone typically loses about a quarter of the weight as lean. In an MRI analysis of 55 people from the STEP UP trial, semaglutide took off 11 litres of fat. Thigh muscle volume fell about 1.1 litres (not statistically clear), fat marbled inside the muscle fell, and people could still do the same number of sit-to-stands as before.²⁶

Two things harden it. The absolute amount is large because the total loss is large: losing 15 kg with a third of it lean is 5 kg of lean tissue. And the people most likely to be prescribed these drugs, older adults and people with diabetes, are the ones who can least afford it. Muscle is where most of the glucose from a meal goes (Exercise, I.2), so losing it works against the very insulin sensitivity the drug is improving.²⁵

V.2Why appetite suppression hits muscle harder

The drug cuts appetite for everything, including protein. People on high doses commonly eat a third to half of what they used to, and meat is often the first thing that stops appealing because it's filling and slow to digest. Muscle is built and kept by two signals, protein intake and loading. A drug that removes most of the first, in someone who wasn't doing the second, is set up to lose muscle. No human trial has measured protein intake on these drugs carefully enough to separate the drug's effect from the diet's.

V.3Lifting decides the outcome

The clearest trial isn't of a drug. In 160 obese adults over 65 who dieted off about 9% of their weight, the aerobic-only group lost 5% of their lean mass; the lifting groups lost 2–3%. Hip bone density fell 3% with aerobic exercise alone and only 0.5–1% with lifting. Strength rose 18–19% in the groups that lifted and 4% in the group that didn't.²⁷ Losing weight while lifting kept most of the muscle and nearly all the bone.

V.4Bone

Weight loss thins bone, partly because bone responds to the load it carries. In the Danish liraglutide trial (IV.4), people on liraglutide alone lost more bone density at the hip and spine than people who exercised, despite similar weight loss. The combination of liraglutide and exercise lost the most weight and kept bone density the same as placebo.²⁸ Bone loss matters beyond fractures, because bone is where the body stores lead (VI.3).

Strongthat a quarter or more of GLP-1 weight loss is lean tissue, and that lifting protects muscle and bone during weight loss; Contestedhow much of the lean loss is real muscle and whether it matters for function.

Fat isn't inert storage. It holds whatever fat-soluble things you've eaten or breathed for years, and it releases them when it's broken down. Slow weight loss releases them slowly. A drug that takes off 15–20% of body weight in a year releases them fast.

VI.1Fat-soluble pollutants

Persistent organic pollutants (POPs) are man-made chemicals that dissolve in fat and resist breakdown: PCBs, dioxins, and old pesticides like DDT and its breakdown product DDE. Everyone carries them, and they're stored mainly in fat. People with obesity carry two to three times the total body burden of lean people.³¹

When fat shrinks, they come out. In 39 obese adults on a 15-week diet who lost an average of 9.5 kg, every one of the 19 organochlorines detected rose in the blood, 15 significantly. The more weight people lost, the more they rose, and levels in the remaining fat rose too, because the same amount of chemical was now packed into less fat.²⁹ In 16 men on a similar diet, the rise in blood pollutants tracked a fall in active thyroid hormone (T3) and a drop in resting metabolism beyond what the weight loss explained.³⁰ That's a correlation, but it points at one possible reason metabolism slows after weight loss (IV.3).

The release is also the way out. After bariatric surgery, blood POP levels rose, and within 6–12 months total body PCBs had fallen 15%. Higher blood POP levels went with worse liver markers, independent of age and weight.³¹ So releasing these chemicals isn't the problem; it's how they leave. The problem is how much is in the blood at once, and whether it's excreted or just moved to the liver and brain.

Most of what the liver excretes into bile gets reabsorbed lower in the gut and goes back around. Breaking that loop speeds removal. In a year-long pilot trial in Anniston, Alabama, a town contaminated by a PCB factory, eating 15 g a day of olestra, a fat that can't be absorbed, made PCBs leave the body about ten times faster than in the years before. Vegetable oil didn't.³² No trial has tested fibre for this in people, but the mechanism is the same one fibre uses on bile acids (Fiber).

VI.2Linoleic acid

Fat tissue records your diet over years. Its linoleic acid content reflects intake over the past one to several years, which is why it's the standard long-term marker of how much seed oil someone eats.³³ In US adults it rose 136% in fifty years (Linoleic Acid). When fat cells break down their stores, they don't release all fatty acids equally: the more double bonds a fatty acid has, the more readily it's released, in animals and in people.³⁴ Linoleic acid, with two double bonds, comes out faster than the saturated and monounsaturated fats around it.

So rapid fat loss sends a large flow of free linoleic acid to the liver and other tissues, at a rate set by how fast the fat is shrinking rather than by what you're eating that day. Linoleic acid is the fat most prone to oxidation, and its oxidised products (OXLAMs and 4-HNE) are the damage the Linoleic Acid paper is about. Part of that flow is simply burned for energy, which is the point of losing fat. There are no human studies measuring whether OXLAMs or other oxidation markers rise during fast GLP-1 weight loss. The physiology says the exposure is there, and that it's larger the more linoleic acid your fat stored and the faster you lose it.

A related point: losing fat removes linoleic acid from the body, but it doesn't change the percentage of linoleic acid in the fat that's left. Only diet does that, slowly. Someone who loses 20 kg on a drug while still eating seed oils keeps fat that's just as high in linoleic acid, and if they regain, the new fat is built from the same diet.

VI.3Heavy metals come from bone, not fat

Heavy metals mostly don't dissolve in fat, so fat loss doesn't release much of them. About 90% of the body's lead is stored in bone; cadmium builds up in the kidney and liver; mercury is spread through blood, kidney and brain. The way weight loss releases lead is by thinning bone (V.4).

One study measured it. In 73 women followed for six months, those who lost 37 kg on a severe diet had higher bone turnover and blood lead rose by about 2 µg/dL. Women who lost 5–6 kg had no change. The more weight lost, the more lead rose, though levels stayed well below what's defined as overexposure.³⁵ The fix is the one for bone: lift, and don't lose weight faster than you need to. Calcium intake was linked to lower blood lead at the start of the study.

VI.4Gallstones

Rapid weight loss makes bile more saturated with cholesterol, and GLP-1 drugs also slow how often the gallbladder empties. Across 76 randomised trials (103,371 people), GLP-1 drugs raised gallbladder and bile duct disease 37%. In the weight-loss trials the risk more than doubled (RR 2.29), and it was highest at high doses and with longer use.³⁶ The gallbladder contracts when fat reaches the gut, so very low-fat eating during weight loss leaves bile sitting longer. Eating some fat at each meal keeps it moving.

VI.5The other side effects

Gut. Nausea, vomiting, diarrhoea and constipation are the most common side effects and usually ease after the dose stops rising.¹¹ In a large insurance-claims database, people prescribed GLP-1 drugs for weight loss had about nine times the rate of pancreatitis, four times the rate of bowel obstruction and nearly four times the rate of gastroparesis (a stomach that barely empties) compared with people on bupropion-naltrexone, another weight-loss drug.³⁷ These are rare events in absolute terms, and the randomised trials haven't shown a clear pancreatitis signal. Because the stomach empties slowly, people on these drugs can still have food in their stomach after a normal fast, which is a risk under anaesthesia. Tell the anaesthetist.

Eyes. At one Boston eye hospital, patients prescribed semaglutide had four to seven times the rate of NAION, a sudden loss of vision from poor blood flow to the optic nerve.³⁸ That's observational and from one centre, but it's a large effect, and sudden vision loss on the drug needs urgent attention.

Hair. Fast weight loss and low protein intake can push many hair follicles into the resting phase at once, causing shedding a few months later (telogen effluvium). It's usually temporary. See Hair.

Face and skin. Fat loss from the face and loose skin elsewhere ("Ozempic face") is what any large fast weight loss does; the drug just makes large, fast loss common.

Strongthat weight loss raises blood POPs and that GLP-1 drugs raise gallbladder disease; Moderatefor lead from bone (one study) and the bowel and eye signals (observational); Inferredfor oxidation from released linoleic acid (physiology; no human studies).

VII.1If you take it

VII.2Raising your own GLP-1

Nothing you eat comes close to the drug's signal (II.3). But your own GLP-1 comes with GLP-2, works through the normal gut-to-brain route, and has none of the side effects.

Protein before carbohydrate. In people with type 2 diabetes, 55 g of whey taken 30 minutes before a carbohydrate meal slowed the stomach, raised GLP-1 the most of the conditions tested, and sharply flattened the glucose spike.³⁹ Eating the meat and vegetables on your plate before the bread or rice does a smaller version of the same thing.

Fermentable fibre. Gut bacteria turn fibre into short-chain fatty acids, and L-cells carry receptors for them; mice lacking those receptors released less GLP-1 and handled glucose worse.⁴⁰ In overweight adults, 10 g a day of a fibre built to deliver propionate to the colon raised GLP-1 after meals, cut how much people ate, and over 24 weeks reduced weight gain, visceral fat and liver fat.⁴¹ Foods that feed the same bacteria include cooked-and-cooled potatoes and rice, green bananas, onions, garlic and beans (Fiber).

Reach the lower gut. L-cells are densest at the far end of the small intestine and in the colon. Whole foods that digest slowly get nutrients down there; refined food is absorbed before it arrives.

Moderatefor whey and propionate-fibre raising GLP-1 (small human trials); Animalfor the short-chain fatty acid receptor mechanism.