AOD-9604 (LAT-8881)

Sixteen residues

AOD-9604 is a sixteen-residue peptide: Tyr-Leu-Arg-Ile-Val-Gln-Cys-Arg-Ser-Val-Glu-Gly-Ser-Cys-Gly-Phe, cyclised by a disulfide bond between the cysteines at positions 7 and 14. Fifteen of those residues are the 177-191 fragment of human growth hormone; the tyrosine at the front is not part of the hormone sequence. PubChem returns a single record for the name - CID 71300630, CAS registry number 221231-10-3, molecular formula C78H123N23O23S2, molecular weight 1815.1 g/mol, InChIKey GVIYUKXRXPXMQM-BPXGDYAESA-N. The FDA Global Substance Registration System carries the same molecule under UNII 7UP768IP4M, with the same registry number, the same sequence and the 7-to-14 disulfide recorded explicitly. Its preferred name on that record is not AOD-9604. It is LAT-8881, which matters further down this page.

Not every identifier in circulation survives checking, and three do not. The description of the compound as hGH 176-191, which appears on vendor product pages and in a 2026 review in Frontiers in Endocrinology, denotes a different molecule: PubChem holds that one separately as CID 172966176, formula C80H127N23O24S2, molecular weight 1859.1 g/mol. The World Anti-Doping Agency lists AOD-9604 and hGH 176-191 as two named examples in one clause, not as synonyms. A CAS number of 386264-39-7 appears in supplier listings and was recorded as such in FDA's 2024 evaluation; a PubChem lookup on it returns nothing. The sponsor's own human safety paper prints a third number, 38624-39-7, which also returns nothing.

What the rodent work measured

The first published animal data came out of Monash University. Ng and colleagues treated obese Zucker rats with an oral dose of 500 micrograms per kilogram daily for 19 days and reported body weight gain of 15.8 plus or minus 0.6 g against 35.6 plus or minus 0.8 g in controls, with raised lipolytic activity in adipose tissue. That report also described euglycaemic clamp measurements in which chronic treatment produced no change in insulin sensitivity, where intact human growth hormone did. Group sizes are not stated in the retrieved abstract. Heffernan and colleagues then treated ob/ob and lean C57BL/6J mice for 14 days by mini-osmotic pump, reporting reduced weight gain, increased in vivo fat oxidation and raised plasma glycerol without the hyperglycaemia that accompanied growth hormone.

The rest of the field rests on two mechanistic observations from that 2001 work. In Ba/F3 cells transfected with the growth hormone receptor, the fragment did not compete with iodine-125-labelled human growth hormone for binding and did not induce proliferation, at concentrations where the intact hormone did both. In a companion paper, 14 days of intraperitoneal treatment in beta-3-adrenergic-receptor knockout mice produced no change in body weight gain, adipose depot size or plasma glycerol, while the same treatment in wild-type obese mice raised beta-3-adrenergic receptor expression; in an acute experiment, however, the fragment still increased energy expenditure and fat oxidation in the knockout animals. Those authors concluded that the lipolytic actions of both human growth hormone and AOD9604 are not mediated directly through the beta-3-adrenergic receptor, although both compounds increase its expression. FDA's 2024 reviewers noted that none of these studies assessed a dose-response relationship and none corrected for multiple comparisons.

One frequently cited 2000 paper in the American Journal of Physiology is about a different peptide altogether: the compound there is AOD-9401, and FDA excluded the paper from its evaluation on that ground. Outside metabolism the animal record is thin. Kwon and Park randomised 32 mature New Zealand white rabbits with collagenase-induced knee osteoarthritis to weekly intra-articular saline, hyaluronic acid, AOD-9604, or the last two combined, in four groups of eight, and reported lower gross morphological and histopathological scores and a shorter lameness period in the combination group.

How the obesity programme ended

Metabolic Pharmaceuticals Limited, an Australian company, ran six trials coded METAOD001 through METAOD006. None was published as an individual paper. They appear in a single 2013 summary by Stier, Vos and Kenley in the Journal of Endocrinology and Metabolism, which PubMed does not index; an author search there returns no record of it. The first four were small - 15, 23, 17 and 36 subjects, intravenous then oral, single doses then seven days. The two that mattered were METAOD005, in which 300 obese adults received 1, 5, 10, 20 or 30 mg orally each day for 12 weeks in groups of fifty, and METAOD006, in which 502 adults were randomised to 0.25, 0.5 or 1 mg daily for 24 weeks. The two available sources disagree on how many were enrolled to reach that number: Stier and colleagues report 534, while the sponsor's exchange filing, as summarised by FDA, reports 536. Both give 502 randomised.

METAOD006, also called the OPTIONS study, was powered for an 80 per cent chance of detecting a 1.8 kg difference from placebo. It did not detect one. On 21 February 2007 the company announced that the phase 2B results did not support commercial viability and that development for obesity was terminated. The 12-week study reached print only as a conference abstract, by Herd and colleagues in Obesity Research in 2005; FDA searched for a full publication of its methods and results and did not find one. That abstract is the origin of a figure still in circulation, that the 1 mg dose produced the largest effect, at a mean rate of 0.22 kg per week against 0.07 kg per week on placebo.

Safety measurements from those trials remain the most substantial human dataset that exists for this compound. In METAOD005, circulating IGF-1 was measured at baseline and at 12 weeks with no significant change in any group, and anti-AOD9604 antibodies were absent at baseline and at 4, 8 and 12 weeks. In METAOD006, mean IGF-1 changes were 1.76 nmol/L at 12 weeks and 1.24 nmol/L at 24 weeks, at p values of 0.51 and 0.76. Five serious adverse events were recorded during active treatment in METAOD005: basal cell carcinoma, squamous cell carcinoma, a lipoma, breast cancer and malignant melanoma. The investigators judged none of them related to study medication.

A second programme, under a different name

FDA's 2024 evaluation states that a search of ClinicalTrials.gov did not list any studies on AOD-9604. That is accurate as written and misleading as read. Three completed randomised trials sit on that registry with results posted, filed under the development code LAT8881 - the preferred name on the same UNII record, 7UP768IP4M. The sponsor is Lateral Pharma Pty Ltd and the indication is not obesity. A 2023 paper in Clinical and Translational Immunology, co-authored by an author affiliated to that company, states plainly that LAT8881 is a synthetic form of the naturally occurring C-terminal fragment of human growth hormone and has been assessed in clinical trials for obesity and neuropathy (Harpur and colleagues 2023, PMID 36969366).

NCT03865953 randomised 53 adults with diabetic peripheral neuropathy or post-herpetic neuralgia to oral capsules twice daily in a two-period crossover. Mean change on an 11-point numeric pain rating scale was minus 0.87 on active treatment and minus 0.74 on placebo, a net difference of minus 0.14 with a 95 per cent confidence interval of minus 0.76 to 0.49 and p equal to 0.67. NCT04153409 enrolled 21 adults treating acute migraine with 60 mg orally at onset; its primary pain score was analysed at seven timepoints between 30 minutes and 24 hours, returning net differences from minus 0.01 to minus 0.89 points at p values from 0.15 to 0.99, none of them significant. The largest of those differences, minus 0.89 at two hours, carried the lowest p value, 0.1488. NCT05298306 gave 26 adults with lumbar radicular pain single intravenous infusions at 0.8, 1.2 and 1.8 mg/kg; net differences across timepoints ran from minus 0.32 to plus 0.32 on a visual analogue scale, at p values from 0.3964 to 0.98, the three-hour figure being posted as a median difference where the remaining timepoints are posted as means.

Serious adverse events numbered zero across all three studies. In the neuropathic pain trial, five participants on active treatment and six on placebo recorded an adverse event above the five per cent reporting threshold, the terms being upper respiratory tract infection and headache. Grade 3 or 4 events numbered zero in every arm of the intravenous study, including the highest dose. Taken with the obesity programme, the human record now stands at four randomised placebo-controlled efficacy comparisons in four different conditions: the weight-loss endpoint of METAOD006 and the pain-score endpoints of NCT03865953, NCT04153409 and the second part of NCT05298306, none of which was met. The registered primary outcome of NCT05298306's first part was adverse events by dose, a safety measure carrying no efficacy hypothesis and no posted analysis.

A binding target, identified in 2026

Reviewers at FDA wrote in November 2024 that the molecular targets for AOD-9604 have not been identified and its mechanisms of action remain unknown. A paper published in Pain in August 2026 speaks to that directly. Spanswick and colleagues used a photoaffinity conjugate of the active metabolite of LAT8881 to pull down lanthionine synthetase C-like protein 1 from spinal cord, then validated the result with siRNA knockdown of LanCL1 in dorsal root ganglion, which abolished the compound's activity. The same paper reported reversal of mechanical allodynia in several rodent neuropathic models, suppression of ectopic firing at the dorsal root ganglion in vivo, and membrane hyperpolarisation in ex vivo spinal cord.

Two qualifications belong beside that finding. The binding was shown for a metabolite rather than the parent peptide, which fits Cox and colleagues' 2015 observation that a nine-residue fragment, CRSVEGSCG, persists in serum considerably longer than the intact molecule. And the paper carries heavy declared interests: two authors are employees and shareholders of the sponsor, two sit on its scientific advisory board, and two hold patents on LanCL1 and its peptide ligands. It is a mechanism for the pain work rather than for the lipolysis work, and the 2001 papers did not leave a lipolytic mechanism standing to compare it with: the chronic effects there required intact beta-3-adrenergic signalling, an acute effect on energy expenditure and fat oxidation survived in the knockout animals, and the authors concluded the lipolytic action is not mediated directly through that receptor.

Two readings of the same toxicology

A single published article carries the entire regulatory toxicology package. Moré and Kenley, writing in 2014 in the same non-indexed journal, describe a four-week intravenous study in Sprague-Dawley rats at 0.1, 1.0 and 10 mg/kg/day with ten animals per sex per group; a six-month oral study in Han Wistar rats at 0.5, 20 and 100 mg/kg/day with twelve per sex per group; and a nine-month oral study in cynomolgus monkeys at 0.5, 10 and 50 mg/kg/day with seven per sex per group. Pharmacokinetics were measured in adult female pigs, where a 400 microgram per kilogram intravenous bolus gave a half-life of roughly three minutes; in rat plasma in vitro the peptide degraded by successive amino-terminal truncation, with a half-life of approximately four minutes and no intact peptide detectable 56 minutes after spiking.

Those authors concluded there were no toxicological concerns. FDA's pharmacology reviewers, reading the same article nine years later, took a different position on the same numbers. At week 13 of the six-month rat study serum osteocalcin fell dose-dependently; at week 26 in females it rose dose-dependently, a pattern the reviewers read as a possible bone-turnover signal. In the monkeys, minimal to slight periportal hepatocyte vacuolation was present in two mid-dose males, in one control female, in one low-dose female and in every female in the high-dose group. Genotoxicity came out equivocal rather than negative: a positive preincubation Ames signal in WP2 uvrA at 200 and 1,000 micrograms per plate that did not reproduce, an increase in bone marrow micronuclei without a clear dose-response, and a chromosome aberration increase at 100 micrograms per millilitre in one run of two.

Several categories of study are missing from the record altogether. No carcinogenicity study, no developmental or reproductive toxicity study, and no repeat-dose study by the subcutaneous or transdermal route has been published or identified. The oral bioavailability figure derived from the pig study came out at 170 per cent, which the authors attributed to variability and FDA read as uninterpretable, several plasma measurements carrying standard deviations approaching the mean. The article provides no data tables, graphs or statistical analyses for any of the repeat-dose studies.

Regulatory position as of August 2026

AOD-9604 is not an approved medicine in the United States and no regulator, registry or pharmacopoeia identified in FDA's search lists it as approved anywhere. On 4 December 2024 the Pharmacy Compounding Advisory Committee considered AOD-9604 free base and AOD-9604 acetate for the section 503A bulk drug substances list, which governs what compounding pharmacies in the United States may lawfully use. The committee voted 12 to nil against inclusion, having agreed 12 to nil to treat the two forms in a single vote. FDA's evaluation found neither substance physically and chemically well characterised and raised immunogenicity concerns specific to the subcutaneous route, for which it located no human exposure data at all.

Elsewhere the picture is consistent. No monograph exists in the United States Pharmacopeia, the European Pharmacopoeia, or the Chinese, Indian or Japanese pharmacopoeias. The 2026 WADA Prohibited List names AOD-9604 at S2.2.3, among growth hormone fragments, prohibited at all times in and out of competition. Australia's Advisory Committee on Medicines Scheduling added it to the Poisons Standard as a performance and image enhancing drug. The compound has been recovered from unlabelled pharmaceutical preparations seized by the Belgian authorities and from vials confiscated in the United States, and a urine assay with a limit of detection of 50 pg/mL has been validated for it.

What is not known

No efficacy endpoint of any trial of this compound has been met. The 24-week obesity trial, the largest ever run, was powered to detect a 1.8 kg difference and did not, and the three pain trials returned net differences of a fraction of a point on their pain scales, with p values across those three records running from 0.15 to 0.99. What that establishes is an absence of demonstrated effect at the doses and durations tested, not an absence of effect. Two of the three pain studies, NCT04153409 and the second part of NCT05298306, are titled proof-of-concept studies; NCT03865953 instead describes itself as a randomised, placebo-controlled, double-blind, crossover, phase IIa study to investigate the efficacy and safety of oral LAT8881 in neuropathic pain, with its design module framing that as testing the concept that the compound is safe and effective in the indication. None of the three registry records retrieved states a power calculation for its efficacy endpoint, so no statement about their statistical power can be sourced. The registered primary outcome of NCT05298306's first part was adverse events by dose, a safety measure rather than an efficacy comparison, and it carries no posted analysis. Nothing is known about the subcutaneous route in humans, which is the route by which the compound is most often encountered outside research: FDA searched for and found no human exposure data for subcutaneous or transdermal administration, and no repeat-dose animal toxicology by either route has been published. The toxicology that does exist rests on one article carrying no data tables and no statistical analyses, whose authors and FDA's reviewers reached opposite conclusions from the same osteocalcin, hepatic and genotoxicity observations - and the hepatic finding included one control female and one low-dose female alongside the mid-dose males and high-dose females, which is exactly the background incidence a reader needs to judge it. There is no carcinogenicity study and no developmental or reproductive toxicity study in any species, which matters because four of the five serious adverse events in the 300-patient trial were malignancies, and because the compound is derived from a hormone whose downstream mediator is implicated in cancer risk. The oral bioavailability figure is uninterpretable, having been calculated at 170 per cent. Immunogenicity was assessed only for the oral route, where no antibodies were detected; the injectable route, where aggregation risk is highest, has never been tested. The LanCL1 binding result was obtained for a metabolite rather than the parent peptide and has not been reproduced by an independent group.

Questions