AOD-9604 vs GLP-1 Peptides

Published August 2026 · Patriot Labs Research Team · ~11 min read

One shelf, two strategies

Shopping categories are a merchandising convenience, not a taxonomy. “Metabolic peptides” groups compounds by the topic researchers associate them with, which is roughly like sorting a toolbox by which room of the house each tool was last used in. Our overview of the metabolic category covers what is on that shelf; this guide is about why two of the items on it have almost nothing in common.

Reduced to one line each:

AOD-9604 is a short synthetic peptide corresponding to the tail end of a large human hormone. The design premise was subtractive — take a molecule with a mixture of desirable and undesirable activities, isolate the region associated with the desirable one, and discard the rest.

GLP-1 receptor agonists, and the multi-receptor agonists that grew out of them, are the opposite. They begin from a fully characterised hormone–receptor pair and ask an engineering question: given a natural ligand that works but survives only minutes in circulation, how is a version built that persists and still activates the receptor?

Subtractive versus constructive. That distinction runs through everything below, including the sizes and shapes of the two literatures.

Strategy one: subtract from a hormone

Human growth hormone is a 191-residue four-helix bundle. It engages its receptor through two distinct binding surfaces, and structural work on the 1:1 complex — solved using an hGH variant carrying a G120R substitution that blocks recruitment of a second receptor copy — showed that a small number of hormone side chains contribute most of the binding energy, packing against a pair of tryptophan residues on the receptor. The point relevant here is that the receptor-engaging machinery of hGH is a specific, mapped set of contacts distributed across the folded protein.

Growth hormone also has a long-recognised association with lipid mobilisation, alongside effects on growth and on carbohydrate handling that a metabolic drug developer would not want. The subtractive idea was to find the part of the molecule responsible for the first and leave the rest behind.

The region chosen was the C-terminus. Published descriptions of AOD-9604 define it precisely: it is the hGH 177–191 region prepared by solid-phase peptide synthesis, carrying an additional tyrosine at the N-terminal end for peptide stabilisation, and cyclised by a disulfide bond between the two cysteine residues. That definition matters commercially, because “hGH fragment 176–191” is sold as though it were the same substance. It is a related but chemically distinct entity, and the human study record described in the literature belongs to AOD-9604.

Two features of this design are worth naming. First, it is a linear-sequence argument: it assumes that a functional property of a folded protein can be carried by a contiguous stretch of that protein once the fold is gone. That is sometimes true for peptide hormones and often is not. Second, the design goal was defined by what the fragment should not do — not raise IGF-1, not disturb carbohydrate handling — more crisply than by what it should bind to. There is no receptor named in the design.

What the AOD-9604 literature actually contains

The primary record is smaller than its commercial footprint suggests, and it is worth walking through in order.

Rodent metabolic work (2000). A study in obese Zucker rats from the Monash group that developed the peptide reported reduced weight gain relative to controls after oral administration over nineteen days, alongside increased lipolytic activity in adipose tissue, and reported no adverse effect on insulin sensitivity as assessed by euglycaemic clamp. The authors concluded the compound “may have the potential to be developed into an orally usable and safe therapeutic agent for obesity.” That is a hypothesis-generating result in a genetically obese rat strain.

Mechanism work (2001). The most informative mechanistic paper compared hGH and AOD-9604 in ob/ob mice and in β₃-adrenergic receptor knockout mice. Both compounds raised the suppressed levels of β₃-AR RNA in obese mice toward levels seen in lean animals. Critically, in knockout mice lacking β₃-AR, chronic treatment produced neither weight loss nor enhanced lipolysis — while acute administration still raised energy expenditure and fat oxidation, which the authors read as evidence of more than one mechanism.

Read carefully, that is a paper describing an indirect mechanism. The chronic effect in that model was contingent on a receptor the fragment is not proposed to bind. Nothing in the sources reviewed here identifies a receptor for AOD-9604 itself.

Human trial programme. Six randomised, double-blind, placebo-controlled trials were run: a Phase I intravenous dose-escalation in healthy males, three Phase IIa studies in obese males, and two Phase IIb studies in obese adults — one over twelve weeks with 300 participants, one over twenty-four weeks with 502. The published account of that programme is explicitly a safety and tolerability paper. It reports no treatment-related withdrawals or serious adverse events, an adverse-event profile comparable to placebo, no detectable anti-AOD-9604 antibodies, no effect on serum IGF-1, and no negative effect on carbohydrate metabolism on oral glucose tolerance testing.

What that publication does not contain is the efficacy result. The trials are described as assessing reduction in body weight alongside safety, but the manuscript presents the safety data only. We were not able to retrieve a peer-reviewed publication reporting the efficacy outcomes of either Phase IIb study. We are not going to fill that gap with inference dressed up as a finding — but the absence itself is information, and so is what happened next.

The repositioning. By 2014 the compound was being written up not as an obesity drug candidate but as a nutraceutical ingredient, in a paper describing non-clinical toxicology and reporting that it had received “generally recognized as safe” status conditional on publication of pre-existing safety data, for intended use in foods, drinks and dietary supplements. By 2015 the published research had moved to an entirely different question: a study in a collagenase-induced knee osteoarthritis model in New Zealand white rabbits examined intra-articular AOD-9604 with and without hyaluronic acid, reporting less cartilage damage and a shorter lameness period than saline, with the combination outperforming either alone. That paper's own background describes AOD-9604 as a peptide that “was originally developed for the treatment of obesity in humans” — past tense.

A compound that completes two Phase IIb trials and then appears in the literature as a food ingredient and a rabbit joint injection did not advance as an obesity therapeutic. That is the honest summary, and it is the single most important thing on this page.

Strategy two: agonise a mapped receptor

The incretin story starts from an observation rather than a molecule. Insulin secretion is substantially higher after oral glucose than after an intravenous load matched for calories — two to three times higher — and this incretin effect has been estimated to account for roughly 50–70% of the postprandial insulin response in healthy individuals. Something released from the gut in response to nutrients is amplifying the pancreatic response.

Two hormones account for most of it. GIP (glucose-dependent insulinotropic polypeptide) is released from enteroendocrine K cells dispersed through the duodenum and jejunum. GLP-1 (glucagon-like peptide-1) is produced by enteroendocrine L cells in the distal gut mucosa of the ileum and colon, generated by cleavage of proglucagon; the same precursor is processed in pancreatic islet α cells and in neurons of the nucleus of the solitary tract in the brainstem, so GLP-1 exists as a gut signal and as a central one.

Both receptors are members of the class B family of seven-transmembrane G-protein-coupled receptors, within the glucagon receptor superfamily. Both ligands engage their own receptor at picomolar half-maximal effective concentrations without cross-reacting with the other. This is the part worth dwelling on: the target is a cloned, named, structurally characterised receptor with measurable potency values, and “receptor agonism” here means something concrete — a ligand binds the receptor and drives the conformational change that couples it to intracellular signalling.

The physiology attributed to GLP-1 receptor activation in the reviewed literature spans several tissues: enhanced insulin secretion, suppression of glucagon release, delayed gastric emptying, and reduced food intake through central appetite pathways. The insulinotropic action is described as glucose-dependent — loss of the effect at basal plasma glucose is what limits hypoglycaemia risk even at high concentrations. The gastric-emptying effect is not permanent under sustained receptor activation; the same literature notes that chronic activation produces a waning effect through tachyphylaxis. GIP behaves differently on that axis: neither acute nor chronic exposure is described as affecting gastric emptying.

Native GLP-1 has one catastrophic property as a drug: it is destroyed almost immediately. Circulating half-life is measured in minutes — one source puts it at one to two minutes under typical physiological conditions — through cleavage by dipeptidyl peptidase-4 (DPP-4). A hormone that works and lasts ninety seconds is a pharmacology problem, not a biology problem, and that is the problem the class was built to solve.

Engineering durability, then adding receptors

The solutions are specific, additive and legible in the sequences themselves.

Block the protease. DPP-4 cleaves at the N-terminus. Semaglutide substitutes the alanine at position 8 with the non-coded residue 2-aminoisobutyric acid (Aib), shielding that site from N-terminal proteolysis. The modification is a direct structural answer to a named enzyme.

Borrow a carrier. Liraglutide and semaglutide both use fatty-acid acylation at lysine 26 to promote non-covalent binding to serum albumin, which slows clearance. A different approach fuses modified GLP-1 sequences to recombinant human albumin outright. Together these push half-life from minutes to hours or days.

Then add receptors. Once one receptor is engineered against successfully, the same peptide backbone can be tuned to engage related class B receptors. Tirzepatide is a dual GIP/GLP-1 receptor agonist. Retatrutide extends this to three: GLP-1R, GIPR and the glucagon receptor. Cryo-electron microscopy structures of retatrutide bound to all three receptors, each coupled to heterotrimeric G_s, were resolved at 2.68 Å, 3.26 Å and 2.84 Å respectively. The peptide adopts a single continuous helix, its N-terminal segment penetrating the receptor transmembrane core while the C-terminal segment engages the extracellular domain. Relative to each receptor's own endogenous ligand it was characterised as roughly 8.9-fold more potent at GIPR and somewhat less potent at GCGR and GLP-1R. Chemically it carries Aib and α-methyl-leucine substitutions and is acylated with a fatty diacid via a linker at lysine 17.

Note what that paragraph contains: three resolved structures, three named receptors, quantified relative potencies, and identified modifications with stated purposes. The within-class differences — single versus dual versus triple agonism — are the subject of our separate comparison of the incretin agonists, and a related but distinct amylin-analogue strategy is covered in the cagrilintide guide, since cagrilintide targets amylin and calcitonin receptors rather than incretin receptors and is often paired with this class in study designs.

The two approaches side by side

The row that carries the argument is the third. One column names receptors and reports binding structures; the other names a mechanism that operates through a receptor the compound is not proposed to bind.

Evidence and regulatory footing

Why the development histories diverged

It is tempting to read the outcome as a verdict on the molecules. A more careful reading is that it is a verdict on two ways of organising a research programme.

A named receptor gives every experiment a readout. When the target is GLP-1R, a chemist who adds Aib at position 8 can ask a bounded question: does the modified peptide still activate the receptor, and does it now resist DPP-4? Each design change is testable against a defined assay before anything reaches an animal. That is why the incretin literature reads as a sequence of solved sub-problems — protease resistance, then clearance, then receptor selectivity, then deliberate multi-receptor engagement.

An activity attributed to a region yields a phenotype and nothing to iterate against. The AOD-9604 programme could measure whether obese rodents gained less weight. What it could not easily do was ask why, in a way that would tell a medicinal chemist which residue to change next. The best mechanistic paper in the set ends up pointing at a receptor the fragment is not proposed to bind, in a knockout model. When the mechanism is indirect and the target unnamed, a disappointing clinical result is also uninformative — there is no hypothesis to revise.

Species translation is where subtractive designs are most exposed. A phenotype in a genetically obese rat strain is a property of that model as much as of the compound. A receptor-level result — this ligand binds this receptor at this potency — carries across species assumptions more robustly, because the receptor can be expressed and tested directly.

None of this means fragment-derived peptides are a dead end as a class; several molecules of research interest are fragments of larger proteins. It means that a fragment without an identified receptor is a harder object to develop, and that the gap between the two columns of the table above is mostly a gap in how much each programme could learn from its own experiments.

What the evidence does not establish

We could not verify the efficacy outcomes of the AOD-9604 Phase IIb trials. The published account of the trial programme presents safety and tolerability only. Claims circulating about specific magnitudes of effect in those studies trace, as far as we could follow them, to company announcements rather than peer-reviewed reports. We have not reproduced them here and readers should treat them as unverified.

The AOD-9604 mechanism is not established at the level vendor copy implies. “Targets fat cells” is a summary of a hypothesis, not a demonstrated binding event. In the sources reviewed, no receptor is identified for the fragment; the chronic effect in the most detailed mouse study required an intact β₃-adrenergic receptor, meaning the pathway described is downstream and indirect. Effects on β₃-AR RNA are transcript-level observations, and transcript changes are not the same as protein or activity changes.

Concentration of the foundational work is a real weakness. The rodent metabolic studies that established the AOD-9604 rationale trace to one institutional group and the company developing the compound. Independent replication by unaffiliated laboratories is sparse in what we located. That is a structural feature of the evidence, not an accusation, and it warrants a discount on everything downstream of it.

Nothing in the incretin literature transfers to AOD-9604, or the reverse. They share a merchandising category and nothing else. Reading receptor-level rigour from one column across to the other is the most common error in material written about this pairing.

The osteoarthritis result is a different question in a different tissue. A rabbit joint model with intra-articular injection speaks to cartilage in rabbits after local administration. It is not evidence for a metabolic mechanism, and it does not rehabilitate the obesity hypothesis.

GRAS status is a food-ingredient determination, not efficacy. The status reported in the literature was described as conditional on publication of pre-existing safety data and concerned intended use in foods, drinks and supplements. It says something about a safety dossier. It says nothing about whether the peptide does what it was designed to do, and it is not a drug approval.

Approvals within the incretin class are not a claim about anything sold here. That approved medicines exist in a mechanism class is a fact about that class's development history. Research-grade peptides are not those medicines, are not manufactured or released to those standards, and nothing in this guide should be read as a comparison of products.

Everything above describes cell, receptor and animal work. Where human trials are mentioned, they are cited for what they reported — in the AOD-9604 case, safety endpoints — and not as evidence of any effect in a person.

Frequently asked questions

Is AOD-9604 a GLP-1 peptide? No, and they are unrelated structurally and mechanistically. AOD-9604 corresponds to the C-terminal region of human growth hormone. GLP-1 agonists are analogues of a proglucagon-derived incretin hormone engineered to activate a class B GPCR. Different parent molecule, different target, different design philosophy.

Is AOD-9604 the same thing as hGH fragment 176–191? Not quite, and the two are routinely conflated. Published descriptions define AOD-9604 as the hGH 177–191 region with an additional N-terminal tyrosine added for stabilisation and a disulfide bond between the two cysteine residues. The unmodified 176–191 sequence is a different chemical entity, and the human study record described in the literature belongs to AOD-9604.

Does AOD-9604 have a known receptor? None that we could identify in the sources reviewed here, and this is the sharpest contrast in the whole comparison. The incretin receptors are cloned, resolved by cryo-EM and characterised by measured potencies. For AOD-9604 the most detailed published mechanism runs through β₃-adrenergic receptor expression in mice rather than through a binding site of its own.

What does “receptor agonist” actually mean in this context? It means the molecule binds a specific receptor and drives the conformational change that activates intracellular signalling, as the natural ligand would. For the incretin receptors this is not a figure of speech — the bound complexes have been imaged and the potencies quantified. Applying the same phrase to a peptide with no identified receptor is imprecise.

Why did the incretin class produce approved drugs when AOD-9604 did not? The most defensible answer concerns how the programmes were structured. A named receptor makes every design change testable against a concrete assay; an activity attributed to a region of a hormone does not. That is an observation about drug discovery, not a claim about what either substance does in a person.

Are these ever studied together? Not meaningfully in what we located. Study designs pairing compounds in this space tend to combine agents with complementary receptor-level rationales — incretin agonists alongside amylin analogues, for instance — rather than pairing a receptor agonist with a fragment whose target is unidentified.

Is it approved for human use? No. AOD-9604 is not an approved drug; it appears on the WADA prohibited list among growth hormone fragments and on the FDA's list of bulk substances nominated for compounding that may present significant safety risks. Approved medicines exist within the incretin class, but the research-grade materials sold here are not those medicines. All Patriot Labs products are sold strictly for in-vitro research and laboratory use only, and are not for human or veterinary consumption.

All Patriot Labs products are sold strictly for in-vitro research and laboratory use only. Not for human or veterinary consumption. This guide is educational and describes peptide chemistry and published research in general terms; it is not medical advice, does not describe how to use any product, and the references cited do not constitute a product claim.