Two numbers are circulating side by side this year. Retatrutide's pivotal obesity trial reported mean weight reduction of -28.3% at 80 weeks. Survodutide's reported -13.0% at 76 weeks. The arithmetic looks trivial: one is roughly twice the other, and the one with three receptors beat the one with two.
It is not trivial, and the receptor count is not the reason. Those figures come from different trials, different populations, different placebo arms and — most consequentially — different statistical estimands. Before any of that, the mechanistic difference between the two compounds is narrower than "three versus two" suggests.
What follows compares retatrutide and survodutide on mechanism, evidence architecture and documentation. Nothing here is a claim about either compound outside the trial populations that generated the data.
The difference is GIP, not the number three
Both compounds are lipidated multi-agonist peptides built on the structurally related glucagon/GLP-1/GIP hormone family. Both engage the GLP-1 receptor and the glucagon receptor (GCGR). That shared pair is the whole of survodutide's pharmacology and two thirds of retatrutide's.
So the difference is not that one adds glucagon. Both add glucagon. It is that retatrutide also engages the GIP receptor and survodutide does not.
That framing matters because it makes the pair look like a clean subtraction — the same GLP-1/glucagon base, with and without a GIP arm. It is not. The two molecules differ in backbone, in acylation, in the internal balance between their shared receptors, and in every parameter of the trials that tested them. A tempting natural experiment sitting in a catalog is still not an experiment anyone ran.
For where GIP agonism sits in the wider receptor-stacking story, see the semaglutide vs tirzepatide comparison and the metabolic class primer.
What the glucagon arm is for, and how anyone knows it is engaged
Glucagon raises hepatic glucose output, the opposite of what a glucose-lowering agent is meant to do. The design argument for putting it back in is that glucagon also increases energy expenditure and acts directly on the liver, and that a sufficiently strong GLP-1 arm offsets the glycemic penalty. Both primary pharmacology papers make that argument with measurements rather than assertion.
Coskun and colleagues (Cell Metabolism 2022;34:1234-1247.e9) reported that in obese mice, retatrutide's weight loss came from adding GCGR-mediated increases in energy expenditure on top of GIPR- and GLP-1R-driven reductions in calorie intake. Two contributions, attributed separately.
Zimmermann and colleagues (Molecular Metabolism 2023;66:101633) did something methodologically neater for survodutide. To show the glucagon receptor was engaged in a living animal rather than only in a well, they used divergent readouts for each arm: glucose tolerance, food intake and gastric emptying for the GLP-1 receptor, and hepatic nicotinamide N-methyltransferase (NNMT) mRNA expression plus circulating amino acids and FGF-21 for the glucagon receptor, supported by knockout and reporter mice.
That NNMT detail is worth pausing on. On this site NNMT has appeared as a drug target — the enzyme 5-Amino-1MQ is designed to inhibit, covered in the NAD+ and methyl-economy roundup. Here it appears as a readout: hepatic NNMT transcription sits downstream of glucagon signaling, so its movement is evidence the receptor was hit. Same enzyme, opposite role, depending entirely on what the study was asking.
Why you cannot rank them on the published potency numbers
Published EC50 values exist for both compounds, and it is tempting to line them up and declare one more glucagon-weighted than the other. Do not. As covered in the receptor pharmacology primer, EC50 is a composite of affinity, efficacy and the assay system. Receptor expression level, cell background, readout and incubation all move it, often by an order of magnitude. The two compounds were characterized in different laboratories in different cell systems for different papers.
What each paper says within its own system is usable. Coskun describes retatrutide as showing balanced GCGR and GLP-1R activity but more GIPR activity — the GIP arm is the strongest of the three, not an add-on. Zimmermann reports survodutide as a potent dual agonist with roughly comparable activity at its two targets in one cell line. Two internally valid statements that do not compose into a ranking.
The estimand is doing more work than the receptor count
Now the headline numbers, read properly.
Retatrutide's TRIUMPH-1 (NCT05929066) enrolled over 2,300 adults with obesity or overweight and at least one weight-related condition, without type 2 diabetes. Lilly's May 21, 2026 topline reported -28.3% at 80 weeks in the highest dose group under the efficacy estimand, against -2.2% placebo. Under the treatment-regimen estimand — which counts the effect of early discontinuation and protocol-prohibited medications — the same group reported -25.0%, against -3.9% placebo.
Survodutide's SYNCHRONIZE-1 was published in the New England Journal of Medicine (2026;395:776-787). Among 725 participants over 76 weeks, the primary analysis used the treatment-regimen estimand and reported -13.0% in the higher dose group against -5.4% placebo.
Three corrections follow.
The placebo arms are not the same. SYNCHRONIZE-1's placebo group lost 5.4% of body weight. TRIUMPH-1's lost 2.2% or 3.9% depending on estimand. Comparing active arms while ignoring a placebo spread of that size compares two different background conditions.
The estimands are not the same. Comparing an efficacy-estimand figure against a treatment-regimen figure is a category error, and it is the comparison that produces the widest gap. Like for like on the treatment-regimen estimand, the figures are -25.0% and -13.0%.
The ordering survives the correction; the margin narrows. Placebo-adjusted and estimand-matched, the separation is real and still large. The point is not that the correction reverses anything. It is that you cannot know it does not until you run it, and the headline pairing gives no way to tell.
The same problem hits the safety columns. SYNCHRONIZE-1 reported gastrointestinal symptoms in 80.9% and 89.7% of the two active groups against 47.9% on placebo. TRIUMPH-1 reported individual events, with nausea reaching 42.4%. An "any GI event" rate and a nausea rate are not the same measurement.
The ordering flips when you change the question
"Which one has better human data" is malformed, because the two programs answer different questions with different maturity.
On liver fat, survodutide is ahead. SYNCHRONIZE-MASLD (Nature Medicine 2026;32:2948-2958) is a completed phase 3 trial in 216 adults with obesity and at-risk metabolic dysfunction-associated steatotic liver disease. It met co-primary endpoints, with 84.2% of treated participants versus 24.3% on placebo achieving at least a 30% reduction in MRI-measured liver fat at 48 weeks under the efficacy estimand. Retatrutide's published liver work is phase 2a. The compound with the smaller weight number has the more mature liver file.
On cardiovascular outcomes, both programs are further along than most of this catalog ever gets, with an instructive asymmetry. Survodutide's SYNCHRONIZE-CVOT (NCT06077864, Boehringer Ingelheim, 5,531 participants across 539 sites) is listed as COMPLETED with a completion date of June 30, 2026, and no results posted. Retatrutide's outcomes trial (NCT06383390, 10,000 participants) remains active and not recruiting. A finished outcomes trial with nothing yet in the literature is exactly the state where secondary sources start filling the gap with expectation.
Retatrutide also has a pre-approval expanded access record on file (NCT07629401, Eli Lilly, posted June 5, 2026, status AVAILABLE). Expanded access supplies an investigational agent outside a trial. It is not an approval and not an efficacy finding.
The comparison nobody has run
A search of ClinicalTrials.gov returns no head-to-head trial of retatrutide against survodutide. What is registered is retatrutide against semaglutide (NCT06260722) and against tirzepatide (NCT06662383), both active and not recruiting, plus an early-phase survodutide comparison against semaglutide (NCT05202353). Each program is measured against the approved incretins, not against the other.
One more registry note, relevant to anyone running this search themselves. A plain interventional search for retatrutide still returns NCT07467447, a "phase 2, randomized, double-blind, placebo-controlled" retatrutide study, currently listed as RECRUITING. Its sponsor is Hudson Biotech, its study identifier imitates Lilly's internal format, it lists a single site, and its FDA-regulated-drug flag is false. It belongs to the batch of fabricated registrations documented in the trial registry literacy post, and unlike some of its siblings it does not declare itself an example. It sorts into the same result list as genuine Lilly trials.
What the documentation should look like
Both compounds are acylated peptides — a fatty-acid chain is attached to extend circulation by promoting albumin binding, the anti-filtration strategy described in the peptidase and clearance explainer. Zimmermann identifies survodutide's as a C18 fatty acid. Both molecules also contain non-proteinogenic residues that no ribosome installs, which forces at minimum a chemical synthesis step.
Four consequences for paperwork.
The theoretical mass must be the fully acylated molecule. A mass calculated from the bare sequence letters will disagree with the product by the mass of the entire lipid chain. This is the same failure mode as a missing acetyl group, scaled up.
Public verification is uneven. Survodutide resolves in PubChem under both its INN and its development code BI 456906 to a single entry, formula C192H289N47O61, roughly 4232 Da. Retatrutide does not resolve in PubChem under either its INN or LY3437943. For that compound, a stated theoretical mass has no convenient independent public cross-check — which makes the presence of an explicit, stated theoretical mass on the documentation more important, not less.
Acylation creates its own impurity class. A des-acyl species differs by a large, easily resolvable mass — but only if the theoretical mass on file is the acylated one. Where more than one attachment site is chemically available, positional acylation isomers are possible: identical mass, different molecule, invisible to intact-mass measurement. Same problem described for PEG conjugates in the MGF and PEG-MGF deep dive.
These are not ordinary reversed-phase analytes. A long lipid chain makes the molecule markedly more hydrophobic than a bare peptide of similar length, and its terminal carboxyl shifts the charge picture described in the charge and solubility primer. Both matter for how a purity number is generated: as covered in reading a chromatogram, species more hydrophobic than the end of the gradient never elute, and are therefore never counted.
FAQ
Does having three receptors make retatrutide the more advanced design? Receptor count is a design choice, not a quality ranking. Adding a receptor adds a mechanism, and also a tolerability surface and a safety question. Neither program was built to answer the count question in isolation.
Which one is better studied? Depends on the question. On weight reduction in obesity, retatrutide's pivotal file is larger and its reported separation from placebo greater. On liver fat, survodutide has a completed phase 3 result where retatrutide has phase 2a. On cardiovascular outcomes, survodutide's trial has finished and retatrutide's has not — though neither has reported.
Can I compare the numbers in two press releases? Not directly. Percent-change figures are only comparable within a matched estimand, a matched duration and a comparable placebo arm. This is structurally the same rule that applies to HPLC purity percentages, EC50 values and published half-lives across this site: the number belongs to its method.
Both sit on the metabolic shelf alongside semaglutide, tirzepatide and cagrilintide. For documentation standards, see /quality/.
This article is educational and for the laboratory research community. Trulogic Labs products are sold for laboratory and research use only and are not for human consumption.