AOD9604: What Happened When a Clean Mechanism Met a Real Trial

AOD9604 was invented in Melbourne. It came out of Monash University and its spin-out Metabolic Pharmaceuticals, built on a specific and testable idea: that the fat-metabolising activity of human growth hormone could be separated from everything else growth hormone does.

It is also one of the few compounds in this catalogue whose central hypothesis was tested properly in humans and did not hold up. That makes it more instructive than most.

The idea: split the molecule, keep one function

Human growth hormone is 191 amino acids and does several unrelated things — it promotes tissue growth via IGF-1, it influences glucose handling in ways that can impair insulin sensitivity, and it has effects on lipid metabolism. Work through the 1980s and 90s suggested these were not inseparable: the lipolytic activity appeared to be concentrated in the C-terminal region, roughly residues 176-191.

AOD9604 is that fragment, with a tyrosine added at the N-terminus for stability. The name is short for "anti-obesity drug 9604". If the hypothesis were right, it would produce hGH's effect on fat without raising IGF-1 and without the glucose effects — a cleaner molecule by design.

In obese mice, it did reduce adiposity. That result is what drove a decade of development.

What the human trials found

The separation part worked. In human studies AOD9604 did not raise IGF-1 and did not produce the effects on insulin sensitivity associated with full-length growth hormone. Mechanistically, the fragment behaved as intended.

The efficacy part did not. A phase 2b trial in obese participants failed to show a significant difference in weight loss against placebo over the treatment period, and development for obesity was discontinued. Metabolic Pharmaceuticals subsequently wound down that programme.

This is a textbook demonstration of the gap between mechanism and outcome. The compound did what the theory said at the level of markers. It did not produce the result those markers were supposed to predict. A rodent finding and a clean pharmacological profile were not enough — and nothing about that was knowable without running the trial.

Why the fragment idea was reasonable

It is worth being fair to the hypothesis, because it was not naive. Splitting a multifunctional hormone into separable domains is a legitimate and productive strategy in pharmacology — it is roughly how selective receptor modulators exist at all. Growth hormone was a good candidate because its unwanted effects are what limit its use, and those effects are mediated substantially through IGF-1, a downstream product rather than a direct action of the whole molecule.

The mechanism by which the C-terminal fragment influences lipid metabolism has never been cleanly settled. Proposed routes have included effects on beta-adrenergic signalling in adipose tissue and on the enzymes governing lipolysis and lipogenesis, but the picture in the literature is not consistent, and the fragment does not act through the growth hormone receptor in the way full-length hGH does. An unresolved mechanism is not itself a mark against a compound — plenty of useful drugs preceded an understanding of how they worked — but it does mean there was never a strong theoretical basis for predicting the size of a clinical effect.

What replaced it

The instructive comparison is with what actually did work in the same indication. AOD9604 was in development through the period when incretin-based compounds were maturing. Those took a different route entirely — not a fragment of an existing hormone repurposed for one of its side activities, but agonists at receptors with well-characterised roles in appetite and glucose handling, developed with large trial programmes attached.

The contrast is not that one mechanism was cleverer. It is that one was tested at a scale capable of detecting whether it worked, and survived that test. See GLP-1 peptides compared and retatrutide vs tirzepatide vs semaglutide for where that line of development went.

Reading a negative result properly

A failed trial is not proof that a compound is inert, and it is worth being precise about what the AOD9604 phase 2b result does and does not establish. It shows that in that population, at those doses, over that duration, against that endpoint, no significant difference from placebo was detected. Different parameters could in principle produce a different result.

What it does establish is that the effect is not large and obvious. Trials of that design are built to detect effects worth having. When one returns nothing, the honest reading is that any real effect is smaller than the effect the study was powered to find — which for a compound marketed on dramatic claims is the relevant point.

Regulatory and anti-doping status

AOD9604 has an unusually tangled status, and anyone handling it in Australia should know it.

It is not an approved medicine in Australia. Attempts to position it as a food or supplement ingredient in other jurisdictions have not succeeded either; the US FDA rejected a new dietary ingredient notification for it in 2015.

Most importantly for Australian readers: AOD9604 appears on the WADA Prohibited List under the peptide hormones and growth factors category. It became widely known here through a high-profile Australian sports supplements investigation. Anyone in a tested sport should treat that status as decisive.

TXLABS supplies AOD9604 strictly as a research reference material for laboratory use only. It is not a medicine, not a supplement, and not for human or veterinary use.

For the family context see how peptides work; for how a fragment's evidence base compares with a full-length analogue in the same axis, see tesamorelin vs CJC-1295.