AOD-9604

AOD-9604 is a sixteen-amino-acid peptide: residues 177 to 191 from the carboxy-terminal helix of human growth hormone, with a tyrosine added at the front for stability and a disulphide bond holding it in a loop. The name stands for anti-obesity drug. It was developed at Monash University and by Metabolic Pharmaceuticals in Melbourne in the late 1990s, on the observation that this region of the hormone carries its fat-mobilising action, to make an obesity treatment that would burn fat the way growth hormone does without the growth-promoting and diabetes-promoting effects that make the whole hormone unusable for that purpose. It is sold today, by subcutaneous injection, for fat loss and increasingly for joints, and it is grouped with the growth-hormone secretagogues in the orthopaedic reviews even though it works nothing like them.

What the fragment cannot do

Growth hormone activates its receptor by pulling two receptor molecules together, using two separate binding sites on the hormone. AOD-9604 overlaps part of one site and has none of the other, so it cannot dimerise the receptor. In competition assays it cannot displace growth hormone, and in a sensitive proliferation test it does not stimulate growth-hormone-responsive cells at any dose. The human trials confirmed what that predicts: across nearly nine hundred participants, serum IGF-1 did not move against placebo at any dose or duration, glucose tolerance did not deteriorate, and no antibodies to the peptide were found.

That is the fragment's whole selling point and also the puzzle at its centre. Growth hormone reaches cartilage through IGF-1, so a fragment that does not raise IGF-1 has no known route to a chondrocyte. The sponsor's own account is that the mechanism is not understood: the peptide shares some effects of the hormone, on fat, and not others, and in obese mice may act partly through adrenergic signalling. The rabbit study below rests its rationale on the sponsor's unpublished finding that the fragment raises proteoglycan and collagen production in cultured bovine chondrocytes, cited to a patent application. Nothing in a journal reports it.

The peptide is also very short-lived. In pigs its half-life after intravenous dosing is about three minutes, with almost nothing left in plasma by twelve; it is chewed down from one end by circulating enzymes, and the two fragments that result keep some of its activity in the dish. Oral dosing produced slower absorption with the same breakdown products, and the sponsor estimates forty percent oral availability from rat radiolabel work, with the peptide spreading to every tissue except the brain.

The obesity programme

Between 2001 and 2006 the sponsor ran six double-blind, placebo-controlled trials in Australia, 893 participants in all: two single-dose intravenous studies in healthy and obese men, two short oral studies in obese men, and two phase 2b trials of daily oral dosing, one of 300 obese adults for 12 weeks at 1 to 30 mg and one of 502 for 24 weeks at 0.25 to 1 mg. What has been published from them is a safety summary, written by a consultant to the sponsor with the sponsor's former medical director and its chief executive as co-authors. It reports no weight result at all. The sponsor's toxicology paper the following year says that weight loss was seen in the early trials and not in the last one, which added a diet and exercise programme, and that an efficacy manuscript was in preparation. Twelve years later none has appeared, and the 2026 reviews record the programme as having failed its primary weight endpoint and been terminated, on the sponsor's own reporting rather than a published trial.

The safety summary is what the orthopaedic literature cites when it calls AOD-9604 safe, and the 2026 review of peptides for surgeons describes it as a meta-analysis of six randomised trials. It is not one: it pools no estimates and appraises nothing, and it is the sponsor's account of its own unpublished trial reports. What it contains is still worth having. The drug was indistinguishable from placebo on adverse events across the oral dose range, with headache and stomach complaints commonest and more frequent at the highest dose; there was no drug-related withdrawal or serious event; and in the 12-week trial five cancers were diagnosed, three skin cancers and a lipoma in the 20 mg group, a breast cancer at 5 mg and a melanoma at 10 mg, none on placebo. The investigator judged them unrelated because none arose in the highest-dose group and the patients had neglected their medical care. IGF-1 did not rise, which removes the expected mechanism, but the judgement compares nothing with the placebo arm.

The rabbit study

Thirty-two young male New Zealand white rabbits had osteoarthritis induced in the right knee by two injections of collagenase three days apart. From the fourth week they received four weekly injections into the joint under ultrasound guidance: saline, high-molecular-weight hyaluronic acid, 0.25 mg of AOD-9604, or both together, eight animals to a group. The dose is the molar equivalent of the 3 mg of growth hormone an earlier rabbit study had used. At nine weeks the saline knees had the worst cartilage on the lateral femoral condyle by both a gross score and a histological one; every treated group scored better; the combination scored better than either agent alone; and the fragment alone did not differ from hyaluronic acid alone. The one number the paper reports is how long the animals limped: 25 days on saline, 15 on hyaluronic acid, 16 on the fragment, 11 on both. The gait observers and the pathologists were blinded.

Read on its own terms, the study says the fragment did what hyaluronic acid did and that the two did more together, in a model of four weeks' treatment and no follow-up. Read against its own text, it has problems its authors do not discuss: the methods give one collagenase dose and the figure legend another, the methods and the legend disagree about when the treatment injections were given, the cartilage scores are shown only as bars without numbers or intervals, the drug came from the sponsor, and no funding source is stated. The authors write that the sample is too small to generalise from. Their discussion also makes the observation on which any long-term use of the growth-hormone axis turns: acromegaly thickens cartilage, and the case for the fragment is that it might repair a joint without doing that. Nobody has tested whether it does either.

Safety and regulation

Under trial conditions, by mouth, for up to six months, the sponsor's data show a profile indistinguishable from placebo, no immunogenicity, and no effect on IGF-1 or glucose. Sponsor-commissioned toxicology found no mutagenicity, no chromosomal damage, and no adverse finding at oral doses of 100 mg per kilogram a day in rats for six months or 50 mg per kilogram a day in monkeys for nine. The 2026 sports-medicine review raises one theoretical concern, that prolonged adrenergic stimulation could disturb autonomic regulation. Nothing has been measured for the subcutaneous route, which no trial used and which is how the peptide is actually taken, and nothing has been measured in a human joint.

AOD-9604 is approved nowhere for any indication. It sits on the FDA's category 2 list of bulk substances for compounding, which one review attributes to immunogenicity concerns although the trials found no antibodies, and the class is prohibited by the World Anti-Doping Agency. The sponsor obtained a "generally recognised as safe" determination for use as a food and supplement ingredient, which is the status the toxicology paper was written to support. This entry is an appraisal of published research, not dosing guidance.

What would move this

A human joint trial of any design would take this to promising, and none exists or is registered. Short of that, three things would change how the rabbit study reads: the two phase 2b obesity trials published in full, so that the safety record can be judged beside the efficacy it came with; the chondrocyte and stem-cell work that motivated the joint study published in a journal rather than a patent; and a second animal study, outside the sponsor's orbit, with cartilage scores reported as numbers. The fragment's short half-life and its lack of a route to IGF-1 are the two facts a mechanism would have to get past, and the rabbit result, if it holds, would be telling us that a molecule can help a joint by a route nobody has yet named.