Pivotal Protocol · Compound Intelligence
Semaglutide, tirzepatide and retatrutide are three answers to the same question. What separates them is how many hormone receptors each one speaks to, and what each added lever buys you.
01 · What GLP-1 actually is
Glucagon-like peptide-1 is an incretin. Cells in the lining of your lower intestine release it the moment food arrives. It does four things: it tells the pancreas to release insulin, but only when blood glucose is already high, which is why low blood sugar is uncommon on this class rather than the standing hazard it is with insulin or sulfonylureas. Uncommon is not never. In the SURPASS-2 trial, blood glucose below 54 mg/dL was recorded in 0.2 to 1.7 percent of people on tirzepatide and 0.4 percent on semaglutide. GLP-1 also suppresses glucagon, cutting the sugar your liver pushes out. It slows how fast your stomach empties. And it signals fullness directly to the brain.
Your own GLP-1 survives about two minutes before an enzyme called DPP-4 cuts it apart. All three drugs here are engineering answers to that one problem: change the amino acids at the cut site so the enzyme cannot grip, then attach a fatty acid chain that binds to albumin in the blood so the molecule circulates instead of clearing. That is how a two minute hormone becomes a weekly injection. Semaglutide also exists as a daily tablet, Rybelsus, which solves the same problem with an absorption enhancer instead.
Two sibling receptors define everything that follows. GIP is the other incretin, acting on fat tissue and on the pancreas. The glucagon receptor is the third lever, and it runs opposite to the other two: it raises the energy you burn at rest and pulls fat out of the liver, but it also raises the sugar the liver releases. That is why glucagon only works as part of a mixture. On its own it would push blood sugar the wrong way.
02 · The three compounds
| Semaglutide | Tirzepatide | Retatrutide | |
|---|---|---|---|
| Receptors | GLP-1 | GIP + GLP-1 | GIP + GLP-1 + glucagon |
| Status | FDA approvedOzempic, Wegovy, Rybelsus | FDA approvedMounjaro, Zepbound | Investigationalnot approved for any use |
| Trial cited | STEP 168 weeks, n = 1,961 | SURMOUNT-172 weeks, n = 2,539 | Phase 248 weeks, n = 338 |
| Weight change, highest dose |
-14.9%2.4 mg vs -2.4% placebo | -20.9%15 mg vs -3.1% placebo | -24.2%12 mg vs -2.1% placebo |
| Cardiovascular outcome data |
SELECT: heart attack, stroke or CV death fell from 8.0% to 6.5%, hazard ratio 0.8017,604 patients, all aged 45 or older with established cardiovascular disease and no diabetes | No dedicated obesity outcome trial reported | None. Too early |
| Source | PMID 33567185, 37952131 | PMID 35658024 | PMID 37366315 |
03 · The only direct comparison
-20.2% vs -13.7% 751 adults with obesity and no diabetes, randomized to the maximum tolerated dose of each drug for 72 weeks. Waist circumference fell 18.4 cm on tirzepatide against 13.0 cm on semaglutide. The trial was open label, meaning participants and investigators knew which drug was being given.
PMID 40353578
-2.30 vs -1.86 Change in HbA1c over 40 weeks in 1,879 people with type 2 diabetes. Tirzepatide beat semaglutide at every dose tested, from -2.01 at 5 mg up to the -2.30 shown here at 15 mg. One caveat that matters: the semaglutide arm used 1 mg, not the 2.4 mg dose used for weight.
PMID 34170647
On weight, tirzepatide wins, and the margin is not small. Two caveats keep it from being the end of the argument: this is a single open label trial, and semaglutide is the only one of the three tested for cardiovascular outcomes in people without diabetes, where it won. Bigger weight loss and fewer heart attacks are related questions, but they are not the same question, and only one of them has been answered for each drug.
04 · The dose response
The headline number from any trial is the top dose result, because that is how a drug wins approval. It is not where the value lives. Both trials published their full dose ladders, and both show the same shape.
| Compound | One third of top dose | Top dose | Share of the top dose result |
|---|---|---|---|
| TirzepatideSURMOUNT-1, 72 wk | -15.0%5 mg | -20.9%15 mg | 72%at one third the dose |
| Retatrutidephase 2, 48 wk | -17.1%4 mg, pooled starting doses | -24.2%12 mg | 71%at one third the dose |
Two independent trials, two different molecules, one answer: a third of the maximum dose delivers roughly 70% of the maximum effect. The dose response curve bends early and then flattens. Side effects do not flatten with it.
Ramp speed is a separate lever from final dose. In the retatrutide trial, gastrointestinal side effects tracked with dose, and the investigators reported them partially mitigated by starting at 2 mg rather than 4 mg. The trial pooled its weight results across both starting doses, so it does not establish that the gentler start reaches the identical endpoint. What it establishes is that the gentler start costs less on the way up.
05 · What the class costs you
A 2026 systematic review in Annals of Internal Medicine pulled together 35 randomized trials that measured body composition rather than weight alone. Researchers set a threshold in advance: if more than about 25% of the weight someone loses comes from muscle related tissue, the loss counts as disproportionate. Across the drug groups the median was 28.3%, with the middle half of studies ranging from 15.9% to 39.9%, and about two thirds crossed that 25% line.
None of that changes what to do. Protein intake and resistance training are the two levers that decide how much of the loss is fat, and neither is optional on this class.
06 · The sleep and melatonin question
Direct answer: there is no evidence that any drug in this class acts on the pineal gland or changes melatonin production. A PubMed search on 2026-08-31 for the pineal gland together with GLP-1, restricted to titles and abstracts, returns two records, and neither studied the drug acting on the gland. A second search for GLP-1 receptor expression in pinealocytes returned nothing at all. There is no receptor localization study, no melatonin suppression finding, and no human melatonin data. That is an absence of evidence, which is not the same as safety having been established.
What does exist points the other way, and at the level of the whole body clock rather than one gland. Two recent reviews describe the incretin system as clock coupled: your own GLP-1 release follows a daily rhythm, that rhythm is blunted in obesity and diabetes, and it is disrupted by shift work, night light and short sleep. The drugs appear to feed back onto central and peripheral clock networks in turn.
The one hard sleep finding is mechanical rather than hormonal: tirzepatide is FDA approved for obstructive sleep apnea. That works by reducing the tissue load on the airway. Better sleep follows from breathing, not from melatonin.
07 · Which one, and when
| If the priority is | The evidence favors | Why |
|---|---|---|
| Proven heart protection | Semaglutide | The only one of the three with a completed cardiovascular outcome trial in people without diabetes, and it won. Note the trial enrolled people who already had cardiovascular disease |
| Weight loss with an approved drug | Tirzepatide | Beat semaglutide head to head on both weight and blood sugar |
| Maximum magnitude | Retatrutide | Highest reported figure of the three, and its weight loss was still trending down at 48 weeks rather than levelling off |
| Certainty and safety record | Semaglutide | The deepest and longest human evidence base by a wide margin |
08 · Sources