A GLP-1 drug does four things that the FDA labels describe in plain terms: it raises insulin when blood sugar is high, lowers glucagon, slows the stomach, and acts on the parts of the brain that regulate appetite. The first two switch themselves off as blood sugar falls. That single property, which the labels call glucose-dependence, explains more about how these drugs behave than any other fact about them.
This page reads the Mechanism of Action and Pharmacodynamics sections of eight FDA labels, pulled from openFDA on 2026-09-17: Ozempic, Rybelsus, Wegovy, Mounjaro, Zepbound, Victoza, Trulicity and Byetta. Where a figure appears it carries its label and section. Nothing here is advice about any treatment; that belongs with a prescriber, as our editorial standards set out.
Why the natural hormone lasts minutes and the drugs last days is a separate story, told on our page about the GLP-1 the body makes itself. This page is about what the signal does once it is there.
The word the labels use: incretin
The oldest label in the set opens its mechanism section with a definition. Byetta (exenatide, NDA021773):
"Incretins, such as glucagon-like peptide-1 (GLP-1), enhance glucose-dependent insulin secretion and exhibit other antihyperglycemic actions following their release into the circulation from the gut."
An incretin is a gut hormone that makes the pancreas respond more strongly to glucose that has been eaten than to the same glucose arriving by a vein. The difference between those two responses is called the incretin effect. GLP-1 is one incretin; GIP is the other, and it matters here because tirzepatide is built to act on both.
The incretin effect is real, and its size depends on the arithmetic
A 1986 study in the Journal of Clinical Endocrinology and Metabolism (Nauck et al, 63(2):492-8) gave six healthy volunteers oral glucose loads of 25, 50 and 100 g, then matched each with an intravenous infusion producing the same blood sugar curve. Whatever insulin the oral load produced over and above the infusion was attributed to incretin factors.
The finding worth knowing is not a single percentage. It is that the estimate moved a long way depending on what was measured:
| Method of estimating the incretin share | Range across the three glucose loads |
|---|---|
| Peripheral insulin concentrations | 74.9% to 93.0% |
| C-peptide responses | 27.6% to 62.9% |
| Calculated insulin secretion rates | 19.2% to 61.0% |
The authors traced part of the gap to the liver, which removed a smaller fraction of insulin after oral glucose (46.9% to 54.6%) than after intravenous glucose (63.4% to 76.5%). So a figure quoted without its method, of the kind "incretins account for 70% of your insulin response", is a number with a hidden choice inside it. Six people is also a small study. What it settles is that the effect exists and grows with the size of the meal; what it does not settle is one number.
Action one and two: insulin up, glucagon down, both conditional
Every GLP-1 label in the set describes the insulin effect as glucose-dependent. Ozempic's section 12.1 says it most plainly:
"Semaglutide reduces blood glucose through a mechanism where it stimulates insulin secretion and lowers glucagon secretion, both in a glucose-dependent manner. Thus, when blood glucose is high, insulin secretion is stimulated and glucagon secretion is inhibited."
Victoza's label adds the other half of the sentence: the insulin secretion "subsides as blood glucose concentrations decrease and approach euglycemia." Byetta's uses almost identical words, and adds that the drug "does not impair the normal glucagon response to hypoglycemia" — the body's own defence against low blood sugar is left working.
The label figures behind these statements are specific. On Ozempic's pharmacodynamic study, 1 mg of semaglutide reduced fasting glucose by 29 mg/dL (22%) and two-hour post-meal glucose by 74 mg/dL (36%) against placebo. On Mounjaro's, 15 mg of tirzepatide reduced fasting glucagon by 28% and the glucagon rise after a mixed meal by 43% at week 28, against no change on placebo.
Where glucose-dependence becomes visible: the sulfonylurea figures
The strongest evidence that the insulin effect is conditional sits in the hypoglycaemia tables, and it comes from what happens when an unconditional drug is added.
Sulfonylureas push the pancreas to release insulin whatever the blood sugar. GLP-1 drugs do not. So if glucose-dependence is real, low blood sugar should be uncommon on a GLP-1 drug alone and should rise sharply when a sulfonylurea is added. The labels show exactly that shape:
| Label, population | Low glucose (<54 mg/dL) |
|---|---|
| Zepbound, type 2 diabetes, with a sulfonylurea | 10.3% |
| Zepbound, type 2 diabetes, without a sulfonylurea | 2.1% |
| Zepbound, type 2 diabetes, placebo | 1.3% |
| Mounjaro 5 / 10 / 15 mg, with a sulfonylurea, up to 104 weeks | 13.8% / 9.9% / 12.8% |
| Zepbound, adults without diabetes (not systematically captured) | 0.3% |
Sources: Zepbound label section 5.7 and 6.1 (NDA217806, effective 2026-08-28); Mounjaro label section 6.1 (NDA215866, effective 2026-07-29).
Two things in that table deserve care. The Mounjaro sulfonylurea figures do not rise with dose — 10 mg is lowest — which is what would be expected if the sulfonylurea, not the tirzepatide dose, is doing most of the work. And the 0.3% figure in people without diabetes comes from a trial the label itself says did not systematically capture hypoglycaemia, so it is a floor rather than a rate.
Every label reviewed carries the same warning — Ozempic 5.4, Mounjaro 5.3, Byetta 5.3 — that combining these drugs with insulin or a sulfonylurea raises the risk of hypoglycaemia, including severe hypoglycaemia, and that the risk may be lowered by reducing the dose of the other drug. Wegovy's and Zepbound's labels go one step further and state that the drug itself "lowers blood glucose and can cause hypoglycemia." Glucose-dependence reduces the risk; it does not remove it. Any change to a diabetes regimen belongs with the prescriber managing it.
Action three: the stomach, and an effect that fades
All eight labels describe slower gastric emptying. They do not describe it with the same weight, and two of them say it wears off.
| Label | How the gastric effect is described |
|---|---|
| Ozempic / Rybelsus | "a minor delay in gastric emptying in the early postprandial phase" |
| Victoza | "a delay of gastric emptying, thereby reducing the rate at which postprandial glucose appears in the circulation" |
| Mounjaro / Zepbound | "The delay is largest after the first dose and this effect diminishes over time." |
| Trulicity | "dose-dependent but is attenuated with adequate dose escalation" |
| Byetta | listed alongside insulin and glucagon as a principal action |
The fading matters for how the drugs are understood. A common explanation of why people eat less on these drugs is that food simply sits in the stomach longer. For tirzepatide the label says the gastric delay shrinks after the first dose, while weight loss continues for many months in the trials, so the stomach alone cannot carry that explanation.
It does not mean the effect disappears. Residual slowing is why anaesthesia societies have published guidance on these drugs before procedures, which is covered on our GLP-1 and surgery page. It is also the most direct route to the nausea, vomiting and constipation that dominate the counted adverse reactions on our side-effects hub.
Action four: the brain
The appetite action is where the labels are most cautious, and where the two semaglutide labels part company.
Wegovy's section 12.1 states that GLP-1 "is a physiological regulator of appetite and caloric intake, and the GLP-1 receptor is present in several areas of the brain involved in appetite regulation," and that "animal studies show that semaglutide distributed to and activated neurons in brain regions involved in regulation of food intake." Section 12.2 adds that semaglutide "decreases calorie intake" and that "the effects are likely mediated by affecting appetite."
Zepbound's label says the same of tirzepatide, and adds that nonclinical studies "suggest the addition of GIP may further contribute to the regulation of food intake."
Note the verbs. The brain evidence the labels cite is from animals; the human statement is that calorie intake falls and that appetite is the likely mediator. That is an honest description of what is established, and it is narrower than much of the coverage that says these drugs "rewire the brain".
One molecule, two labels, two stories
Ozempic and Wegovy both contain semaglutide, and the first sentence of each label's mechanism section is word for word the same: a GLP-1 analogue with 94% sequence homology to human GLP-1. After that they diverge.
| Ozempic (type 2 diabetes) | Wegovy (weight management, cardiovascular risk) | |
|---|---|---|
| What comes next in 12.1 | Albumin binding, DPP-4 resistance, then glucose-dependent insulin and glucagon | Appetite, caloric intake, brain regions |
| Gastric emptying | "minor delay… early postprandial phase" | "Semaglutide delays gastric emptying" |
| An admitted unknown | — | "The exact mechanism of cardiovascular risk reduction has not been established." |
Labels are written for the indication they carry, so each foregrounds the action that explains the outcome it was approved on. That is reasonable, and it is also a reminder that "how GLP-1 works" has no single answer independent of what the drug is being used for. A reader comparing the two labels could conclude the drug works mainly through blood sugar or mainly through appetite, depending on which one they opened. Both descriptions are the manufacturer's, and both are true of the same molecule.
The Wegovy sentence about cardiovascular risk is worth reading twice. The drug carries an approved cardiovascular indication, and its own label says the mechanism behind that benefit is not known.
Tirzepatide: two incretins, not one
The Mounjaro and Zepbound labels describe tirzepatide as "a GIP receptor and GLP-1 receptor agonist" that "selectively binds to and activates both the GIP and GLP-1 receptors." The other drugs in the table above act on the GLP-1 receptor only, with the sequence homology each label states: liraglutide 97%, semaglutide 94%, dulaglutide 90%; exenatide's sequence "partially overlaps" human GLP-1.
Mounjaro's label also reports an action the GLP-1-only labels do not headline: tirzepatide "increases insulin sensitivity, as demonstrated in a hyperinsulinemic euglycemic clamp study" after 28 weeks. Which receptor produces that is not separated out in the label. The register of which drug acts on which receptor, with approval status, is on our page about which drugs are GLP-1s.
Why the drugs last and the hormone does not
The mechanism sections also say why a weekly injection is possible. Victoza's label states that the body's own GLP-1(7-37) has a half-life of 1.5 to 2 minutes, cleared by the enzymes DPP-4 and neutral endopeptidase. Semaglutide is "stabilized against degradation by the DPP-4 enzyme" and protracted by albumin binding; tirzepatide carries "a C20 fatty diacid that enables albumin binding." The full comparison, and what it means for foods and supplements that claim to raise GLP-1, is on our page on food, supplements and endogenous GLP-1. For the separate question of why the same molecule needs roughly seventy times as much drug when swallowed, see GLP-1 pill vs injection.
What the labels do not say
- No label quantifies how much of the weight loss comes from each action. Appetite is called the likely mediator; the split between brain, stomach and anything else is not stated.
- The brain evidence cited in the labels is from animal studies. The human finding is reduced calorie intake.
- Glucose-dependence lowers hypoglycaemia risk without removing it. Both weight-management labels state the drug itself can cause it.
- The cardiovascular mechanism is stated as not established on Wegovy's label.
How this page was built
Label sections 5 (Warnings and Precautions), 6 (Adverse Reactions), 12.1 (Mechanism of Action) and 12.2 (Pharmacodynamics) were retrieved from the openFDA drug-label endpoint on 2026-09-17, one current label per brand, and quoted with their application numbers and effective dates. The 1986 incretin study was read from its PubMed abstract (PMID 3522621). Where a label describes an animal finding, this page says so.