What it is
AOD-9604 is a synthetic peptide of 16 amino acids, cyclised by a disulfide bond between its two cysteines, taken from the tail end of human growth hormone. Its one letter sequence is YLRIVQCRSVEGSCGF. It came out of Monash University in Melbourne in the 1990s, was developed by Metabolic Pharmaceuticals as an oral obesity tablet, and was later renamed LAT8881 by Lateral Pharma and turned toward pain. It has never been approved as a medicine anywhere.
The description repeated almost everywhere is wrong. AOD-9604 is usually called hGH 176-191 with a tyrosine added to the N-terminus. Mature growth hormone residue 176 is phenylalanine, per UniProt P01241, so hGH(176-191) is FLRIVQCRSVEGSCGF, already 16 residues. AOD-9604 is also 16 residues. It is hGH(176-191) with the phenylalanine at 176 substituted by tyrosine, which is the same statement as hGH(177-191) with a tyrosine added at the front. Adding a tyrosine to the 16-mer would give a 17-mer, YFLRIVQCRSVEGSCGF, which nobody has studied or sold.
The formulae settle it. AOD-9604 is C78H123N23O23S2 and hGH(176-191) is C78H123N23O22S2, differing by a single oxygen atom, the hydroxyl that distinguishes tyrosine from phenylalanine, a mass difference of 16.0 daltons. Appending a whole tyrosine would add about 163.2 daltons. The primary literature uses the 177-191 convention: Cox 2014 describes amino acids 177-191 with an additional N-terminal tyrosine, and the sponsor’s own review calls it Tyr-hGH177-191.
This is not pedantry. HGH Fragment 176-191 is sold separately, is a different molecule one hydroxyl away, and has essentially no human trial record of its own. Almost all human data attributed to it is in fact AOD-9604 data.
Mechanism, and where the species boundary sits
The mechanism given on most pages is the beta-3 adrenergic receptor. The paper cited for it concluded the opposite. Heffernan 2001 dosed obese mice and beta-3 receptor knockout mice for 14 days and reported, in its own abstract, that the lipolytic actions of both growth hormone and AOD9604 are not mediated directly through the beta-3 receptor, although both raise its expression. That is a partial negative, not a positive.
It also does not transfer across species. Human fat cells carry far fewer beta-3 receptors than mouse fat cells, and the selective human beta-3 agonist L-796568 failed to reduce weight or raise 24 hour energy expenditure in obese men over 28 days, per Kennett and Clifton 2010. Gary Wittert of the University of Adelaide, who ran five of the six human studies, told ABC Science in 2013 that the drug does not work in people because “the receptor on which it appears to be dependent is not the same in mice and humans,” adding: “I cannot understand why athletes are taking it.”
The current mechanism candidate is narrower. Spanswick 2026 pulled down LanCL1, lanthionine synthetase C-like protein 1, as the binding target of the active metabolite in rat spinal cord, validated by siRNA knockdown in dorsal root ganglion. Harpur 2023 reports the peptide acts independently of the growth hormone receptor, so legacy database annotations naming that receptor are out of date. Both findings are rodent and in vitro, both sit in a pain context, and neither has anything to do with fat.
One pharmacokinetic figure is worth carrying. In pigs the intravenous half-life was around 3 minutes, per Moré and Kenley 2014, while oral dosing gave an area under the curve 8.5 times higher. The molecule was developed as a tablet for a reason.
What the six human trials actually did
Metabolic Pharmaceuticals ran six randomised, double-blind, placebo-controlled studies, METAOD001 to METAOD006, in roughly 900 people. Four were oral. Two were intravenous. None was subcutaneous. FDA states the gap in its own words: the nominations “did not include, and we did not identify human exposure data on this substance administered via two of the proposed routes of administration (ROAs), SC or topical.” The first four were safety and pharmacokinetic studies at single intravenous doses of 25 to 400 micrograms per kilogram and oral doses of 9 to 54 mg.
METAOD005 produced the famous number. 300 adults with obesity, oral once daily, arms of 1, 5, 10, 20 and 30 mg against placebo, 12 weeks. Kennett and Clifton 2010, the peer-reviewed source, reports mean weight loss of 2.6 kg in the 1 mg arm against 0.8 kg on placebo, a difference of 1.8 kg. The sponsor announcement of December 2004 gave 2.8 kg against the same 0.8 kg. Both figures are real, and the peer-reviewed one is the higher tier.
The qualifier that never travels with the number is that the dose response was inverted. The lowest dose produced the largest effect and the higher doses less, the 10 mg arm showing a smaller reduction than the 1 mg arm. A monotonic dose response is one of the standard checks on whether an effect is real, and this one failed it. The follow-on study then tested 0.25, 0.5 and 1 mg.
That follow-on study, METAOD006, named OPTIONS, was the confirmatory one: 536 enrolled and 502 randomised adults with obesity, oral tablets once daily, 24 weeks, primary endpoint weight loss at 12 weeks across the three dose levels against placebo. It failed. FDA’s 2024 review states the study “enrolled 536 patients with obesity and did not find a significant difference in weight loss,” and quotes the developer saying the results “do not support the commercial viability of the drug as a treatment for obesity.” Wittert, on the six month trial: “quite clearly there was no effect on weight.” The programme was terminated in 2007.
No arm-level kilogram figures for the 24 week study have ever been published: not in the sponsor safety paper, the sponsor review, the Australian registry record or FDA’s briefing document. Any page quoting a weight change for that study is quoting something that does not exist. Across the programme, the number of human trials in which AOD-9604 met a prespecified efficacy endpoint is zero.
The trials nobody connects to it
Lateral Pharma acquired the asset in 2015, renamed it LAT8881, abandoned obesity and redeveloped it for pain. Three registered trials ran under that name and they are the only AOD-9604 human trials with NCT numbers. NCT03865953, a phase 2a oral crossover in 53 people with painful diabetic neuropathy or post-herpetic neuralgia. NCT04153409, an oral proof of concept in 21 people with episodic migraine. NCT05298306, a two part phase 1 in 26 people, intravenous, in lumbar radicular pain.
All three are marked completed with results posted, but those posted numbers could not be verified for this entry, so no outcome figure is attributed to them here. Nothing was published in the peer-reviewed literature, neither indication was carried forward, and the company’s plan as of September 2025 is to start again with a shorter six amino acid peptide, LAT9997.
Safety, and the phrase to be careful with
The consolidated safety publication, Stier 2013, concludes that AOD9604 “displayed a very good safety and tolerability profile indistinguishable from placebo.” That sentence is the sponsor’s. It sits in a journal not indexed in PubMed or MEDLINE, is co-authored by an employee of Metabolic Pharmaceuticals, and was published to satisfy a condition attached to the company’s own GRAS determination. It is also the only consolidated account of the six trials, so it is indispensable and low tier at once.
The underlying numbers are less smooth. Adverse event rates were high in every arm, including placebo: 88.9 percent during the treatment phase of the 12 week study and 78.7 percent in the 24 week study, where the arm by arm range of 75.6 to 83.2 percent does genuinely overlap placebo. Headache dominated the intravenous studies, reported by 16 of 23 subjects in one. Gastrointestinal events rose at the 54 mg oral dose.
The 12 week study recorded five serious adverse events among 300 people, all neoplasms: basal cell carcinoma, squamous cell carcinoma, melanoma, breast cancer and a lipoma. The investigators judged none treatment related, and among adults with a body mass index of 35 or above some incidental diagnoses over 12 weeks are expected. That judgement is probably right, and the cluster still belongs in a safety summary.
On immunogenicity, no anti-AOD9604 antibodies were detected in the subsets assayed. The subsets were selected rather than complete, the routes oral and intravenous, the longest exposure 24 weeks. FDA has separately written that it “is concerned about the potential for immunogenicity of AOD-9604 (free base) when formulated in an injectable dosage form for SC administration due to the potential for aggregation.” Non-immunogenic overstates what was measured: no antibodies were found by the routes tested, and the route people use was never tested.
Regulatory status, precisely
There is no approval anywhere. WADA describes it as a substance still under pre-clinical and clinical development that “has not been approved for therapeutic use by any government health authority.” It is prohibited in sport at all times, at section S2.2.3 among growth hormone fragments, with hGH 176-191 beside it as a separate substance.
Calling it FDA approved, or FDA GRAS, is a category error twice over. What exists is a self-affirmed conditional GRAS determination announced in June 2012 by a privately convened panel, for an intended oral food use of up to 1 mg per day, conditional on publishing the pre-existing safety data. FDA grants nothing in that process, and neither a GRN number nor a new dietary ingredient notification was located, which are checked-and-not-found negatives rather than proven absences. A food determination at 1 mg a day by mouth is not a safety finding for an injection.
FDA’s written position is dated 5 November 2024: “we believe the evaluation criteria weigh against placing AOD-9604 (free base) or AOD-9604 acetate on the list of bulk drug substances.” That briefing document also finds the free base “not physically and chemically well characterized,” states that “there is a lack of evidence to support the effectiveness of AOD-9604 (free base) and AOD-9604 acetate for the treatment of obesity,” and cautions that the absence of adverse event reports “does not imply that the substance is safe or lacks toxicities.” The advisory committee is reported to have rejected it on 4 December 2024, a fact carried only by an advocacy group’s account with no vote counts located, and it now sits on FDA’s list of substances nominated but withdrawn.
What the record gets wrong
The human fat cell result belongs to another molecule. Heffernan 2000 is cited constantly for the claim that AOD-9604 increases lipolysis in human adipose tissue. That paper is about AOD-9401, a different and earlier Metabolic analogue, and says so in its title and abstract. No human tissue study of AOD-9604 exists.
The founding paper argues against the headline claim. Wu and Ng 1993, titled for the antilipogenic action of the synthetic 177-191 sequence, reported that “no significant lipolytic effect of hGH 177-191 was found” in rat fat pads and concluded the fragment acts on lipogenesis rather than lipolysis. The oldest paper in the lineage says close to the opposite of the framing built on it.
Glucose safety is being read as efficacy. The human trials measured glucose tolerance, insulin sensitivity and IGF-1 and found no adverse effect, and no effect on IGF-1 at all. That is a real and well supported safety finding: the peptide avoids what growth hormone does. It is not evidence that it does anything to fat, which the same programme twice failed to show.
And the dose is the wrong route from the wrong study. The 0.25, 0.5 and 1 mg daily figures are genuine, and they were oral tablets in the 24 week trial that missed its primary endpoint. They are not injection doses, no human has received this peptide subcutaneously in any trial, and the most repeated number on the open web is a failed trial’s tablet dose wearing a syringe.