GLP-1 and Incretin Drugs, Explained

GLP-1 drugs have moved from a niche diabetes treatment to one of the most discussed medication classes in the world. The names pile up quickly: Ozempic, Wegovy, Mounjaro, Zepbound, plus a growing list of investigational compounds with names like retatrutide and CagriSema. This guide steps back from the brand names to explain the underlying biology, how the different drug generations actually differ, what the clinical trials have shown, and what is known about the risks. It is educational only. It contains no dosing information and no guidance on obtaining these drugs.

The incretin system

The starting point is a quirk of human physiology called the incretin effect. When you swallow glucose, your body releases far more insulin than it does when the same amount of glucose is delivered intravenously. That difference accounts for roughly 50-70% of the insulin your body secretes after a meal in healthy people, and it is noticeably blunted in type 2 diabetes.

Two gut hormones drive this effect. GIP (glucose-dependent insulinotropic polypeptide) and GLP-1 (glucagon-like peptide-1) are released from cells in the intestinal lining after you eat. Modern incretin drugs are engineered to mimic or modulate these hormones, sometimes alongside two related signals, glucagon and amylin.

GLP-1 acts on receptors on the insulin-producing beta cells of the pancreas. Its effects are worth listing because they explain both the benefits and the side effects of the whole class: it stimulates insulin secretion, suppresses glucagon, slows the rate at which the stomach empties, and acts on appetite and satiety centers in the brain. A key safety nuance is that GLP-1’s insulin effect is glucose-dependent — it only pushes insulin when blood sugar is already elevated. That is why GLP-1 drugs carry a relatively low intrinsic risk of hypoglycemia compared with older diabetes drugs.

GIP is also insulinotropic, and here the science contains a genuine, unresolved debate worth flagging honestly. Both agonism (switching the receptor on) and antagonism (blocking it) appear to aid weight loss when paired with GLP-1 activity. Tirzepatide is a full GIP receptor agonist, while MariTide (an investigational Amgen compound) is a GIP receptor antagonist combined with a GLP-1 agonist. This contradiction was the subject of a 2025 point-counterpoint exchange in the journal Diabetes, and it has not been settled.

Two further hormones round out the class. Glucagon normally raises blood sugar, but agonizing its receptor also increases energy expenditure and helps mobilize liver fat — which is why GLP-1/glucagon and triple agonists target it, balancing that against GLP-1’s glucose-lowering. Amylin, co-secreted with insulin, contributes to satiety and slowed gastric emptying; its mechanism is already validated in humans by the approved analog pramlintide.

The drug generations

It helps to think of these drugs in generations defined by which receptors they engage:

First-generation GLP-1: liraglutide (Victoza, Saxenda), a daily injection.
Second-generation GLP-1: semaglutide (Ozempic, Wegovy, Rybelsus), weekly injection plus an oral form. An oral version (Wegovy 25 mg) was approved in late December 2025.
Dual incretin (GIP + GLP-1): tirzepatide (Mounjaro, Zepbound), weekly.
GLP-1/glucagon duals: survodutide, mazdutide, and pemvidutide — investigational in the US. Mazdutide is approved in China for weight management and type 2 diabetes.
Triple agonist (GIP + GLP-1 + glucagon): retatrutide, in phase 3 trials.
Amylin combination: cagrilintide plus semaglutide (CagriSema), investigational and not FDA-approved as of this writing.

What the trials show: weight and glucose

The efficacy data here come from large randomized controlled trials, and the numbers have climbed with each generation.

Liraglutide in the SCALE trial (56 weeks, 3,731 participants without diabetes) produced a mean loss of about 8.4 kg versus 2.8 kg on placebo.

Semaglutide in STEP-1 (68 weeks, 1,961 participants without diabetes) produced a mean reduction of 14.9%, versus 2.4% on placebo, with 86% of participants losing at least 5% of body weight. The oral 25 mg form, studied in OASIS-4, showed roughly 16.6% versus 2.7% under a full-adherence analysis, and about 14% under the more conservative treatment-policy analysis.

Tirzepatide in SURMOUNT-1 (72 weeks) produced reductions of 16.0%, 21.4%, and 22.5% at its three doses, versus 2.4% on placebo. In SURMOUNT-5, a head-to-head trial, tirzepatide reached 20.2% versus semaglutide’s 13.7% — though this was a manufacturer-funded, open-label study, which is worth keeping in mind.

The investigational compounds push further but rest on thinner evidence. Retatrutide in a phase 2 trial (48 weeks, 338 participants) reached roughly 22-24% at its higher doses — promising, but phase 2 is early relative to the approved drugs. CagriSema in REDEFINE-1 reached 20.4% versus 3.0% on placebo (treatment-policy), or 22.7% with full adherence. Survodutide, aimed at liver disease (MASH), improved disease without worsening fibrosis in 47%, 62%, and 43% of participants across its three doses versus 14% on placebo — a notably non-monotonic result, meaning the highest dose did not give the best response.

What the trials show: cardiovascular outcomes

Some of the strongest evidence for this class concerns heart outcomes, not just weight.

LEADER (liraglutide, high-CV-risk type 2 diabetes, 9,340 participants) found about a 13% relative reduction in major adverse cardiovascular events.
SUSTAIN-6 (semaglutide) found a hazard ratio of 0.74, though as a pre-approval safety trial its superiority signal was nominal rather than a primary endpoint.
SELECT (semaglutide in people with obesity and established cardiovascular disease but without diabetes, 17,604 participants) found a hazard ratio of 0.80 — the first cardiovascular-outcome win in obesity without diabetes.
SURPASS-CVOT (tirzepatide versus dulaglutide) was an active-comparator non-inferiority trial, with a hazard ratio of 0.92.

The honest framing: the GLP-1 cardiovascular benefit is robust and replicated against placebo. Tirzepatide’s cardiovascular data, by contrast, were generated against another active drug — so the conclusion is “at least as good as a proven medication,” not a fresh demonstration of beating placebo.

Side effects and risks

The most common side effects are gastrointestinal: nausea, vomiting, diarrhea, and constipation. These are dose-related, usually mild to moderate, often transient, and are the leading reason people discontinue. Gradual dose escalation is the standard way to limit them.

Several more serious risks deserve attention:

Thyroid C-cell tumors (boxed warning): Based on rodent studies; human causality is not established. These drugs are contraindicated in people with a personal or family history of medullary thyroid carcinoma or MEN2. Reassuringly, a large 2025 human cohort study found no increased thyroid-tumor risk.
Pancreatitis: Labels instruct stopping the drug if it is suspected.
Gallbladder disease: Modestly increased, largely a consequence of rapid weight loss itself.
NAION (non-arteritic anterior ischemic optic neuropathy): An emerging and debated eye signal. A 2024 cohort study and a Danish-Norwegian cohort (pooled adjusted hazard ratio 2.81) reported an association, but the absolute risk is small (on the order of a few excess events per 10,000 person-years), the studies are retrospective, and causality is unproven. This should be read cautiously.
Other label items: delayed gastric emptying (relevant to anesthesia and aspiration risk), increased heart rate, loss of lean mass with rapid weight loss, and weight regain after stopping.

The approved versus compounded landscape

Through 2024 the FDA listed these drugs as in shortage, which temporarily allowed compounding pharmacies to produce versions of them. That window has closed. The tirzepatide shortage was resolved in December 2024 and the semaglutide injection shortage in February 2025; the FDA’s enforcement discretion for compounders ended on a staggered schedule through mid-2025. Compounding pharmacies (whether state-licensed 503A pharmacies or registered 503B outsourcing facilities) are now barred from making “essentially copies” of these approved drugs.

In 2026 the FDA went further, proposing (announced April 30, 2026) to permanently exclude semaglutide, tirzepatide, and liraglutide from the 503B bulk-substance list on a finding of no clinical need, with public comment open through late June 2026. As of this writing that is a proposed rule, not finalized.

The FDA has also publicly flagged concerns about unapproved or “research-grade” GLP-1 products: dosing errors from self-measured material, unverified salt forms (semaglutide sodium or acetate are not the approved active ingredient), and counterfeit or illegally imported ingredients. These concerns are directly relevant to anyone evaluating product quality. Our how to read a COA guide explains what a certificate of analysis can and cannot tell you, and our reconstitution and handling page covers the practical hazards the FDA raises around self-measured product.

One note for athletes: GLP-1 receptor agonists are not on the World Anti-Doping Agency’s 2026 Prohibited List, but they are on WADA’s Monitoring Program, and detection methods are under active development. “Monitored,” not “cleared,” is the accurate status.

Bottom line

The incretin drugs are one of the best-evidenced medication classes in modern metabolic medicine, with replicated weight, glycemic, and (for several agents) cardiovascular benefits from large randomized trials. The science is also still moving: the GIP agonism-versus-antagonism question is genuinely unsettled, the triple agonists and amylin combinations rest on earlier-stage data, and risk signals like NAION remain under study. Two practical realities frame the current moment — the compounding pathway has largely closed, and the FDA has been explicit about the quality and safety risks of unapproved versions. None of the above is medical advice or a recommendation to use, obtain, or self-measure these compounds; it is a map of what the published evidence currently supports.