GLP-1 + AOD-9604: the case that doesn’t quite close.

Why this question is everywhere

Search volume for “Ozempic face” and “GLP-1 muscle loss” doesn’t fall. It climbs every month. The reason is straightforward: people on semaglutide and tirzepatide are losing real weight, looking in the mirror, and seeing a face that doesn’t look like recovery. It looks deflated. The gluteofemoral area holds; the cheeks, temples, jawline don’t. Some weight came off muscle. Some came off the fat pads that give a face its structure. The deficit found whatever it could find.

Into that gap walks a peptide. AOD-9604, sometimes labeled HGH Fragment 176-191, gets sold as the answer — a synthetic piece of growth hormone that supposedly burns fat without touching muscle, without raising IGF-1, without the cancer risk of full HGH. On paper, it sounds like exactly what a GLP-1 user needs.

The story falls apart in stages. This article walks through them.

The lean-mass cost, in actual numbers

The cost is real and it’s well-documented. Two trials matter here.

STEP 1 (semaglutide). A DEXA-based body composition substudy of about 140 participants from the 68-week STEP 1 trial of semaglutide 2.4 mg weekly showed roughly a 9.7% absolute reduction in total lean body mass over the course ³⁴. Novo Nordisk’s public framing emphasizes that lean as a proportion of total mass actually improved by about three percentage points. That’s technically true and not the argument users care about. In absolute terms, the average participant lost five to six kilograms of lean tissue alongside their fat. That is the body composition signal.

SURMOUNT-1 (tirzepatide). A DEXA substudy of about 160 SURMOUNT-1 participants at 72 weeks documented a 21.3% total weight reduction with the loss distributing roughly 75% as fat and 25% as lean — about a 10.9% reduction in lean mass in absolute terms ⁵. Tirzepatide is being marketed as having a more favorable composition ratio than semaglutide. That’s defensible. It still represents substantial lean loss across long maintenance courses.

The 2024 Diabetes Care review by Locatelli and colleagues characterizes lean-mass losses on incretin therapy as approximately ten percent or roughly six kilograms — “comparable to a decade or more of aging” — and recommends supervised resistance training as a default adjunct ⁶. That is the evidentiary floor of this conversation.

The visible face change has its own anatomy. The plastic surgery literature describes “Ozempic face” as a real composite: rapid subcutaneous fat reduction across the malar, buccal, and temporal compartments, mild masseter atrophy, and thinning of the dermal collagen-elastin support ⁷⁸. The mirror is not lying. The question is what to do about it.

“A drug that drives a sustained caloric deficit doesn’t choose what gets burned. The body chooses, and it never chooses the way the mirror wants.”

What AOD and Fragment are supposed to do

The mechanism story for AOD-9604 and Fragment 176-191 is the cleanest part of this article, so it’s worth getting right before we get to where it stops.

Both compounds are short peptides derived from the C-terminal end of human growth hormone — specifically the residues 176 to 191 of the full hGH molecule. AOD-9604 adds a tyrosine to the N-terminus for stability (Tyr-hGH 177–191). Fragment 176-191 is the same core sequence without that tyrosine. Functionally similar; commercially often interchanged. Both were engineered on the same hypothesis: that you can isolate the lipolytic activity of growth hormone from its growth-promoting and insulin-antagonist activity by using only the fragment that pulls the fat signal.

The mechanistic claim has support in the mouse work. Heffernan and colleagues at Monash showed that AOD-9604 reduced body weight and body fat in obese mice over a two-week course, partly by upregulating β3-adrenergic receptor expression in adipose tissue ¹. Subsequent preclinical and Phase 1 review work documented that, unlike full-length GH, the 176-191 fragment did not produce measurable IGF-1 elevation ². That IGF-1-neutral piece is the entire pharmacological reason this molecule was developed. If it raised IGF-1, it would be a worse version of growth hormone and there would be no reason to make it.

On the GLP-1 side, the mechanism is just as clean: incretin agonism amplifies glucose-dependent insulin secretion, slows gastric emptying, and acts on hypothalamic appetite circuits to cut energy intake. The result is a sustained caloric deficit. The drug does not preferentially burn fat — it tells the brain you’re full and the body liberates whatever substrate it can.

The combination thesis is therefore: the GLP-1 creates the deficit, the AOD/Fragment biases substrate use toward fat. Two compounds, two roles, two non-overlapping mechanisms. On a whiteboard, this looks ideal. On a whiteboard.

The trial record (it’s short)

This is the load-bearing section. Most articles you’ll find on AOD-9604 skip it.

AOD-9604 was developed by Metabolic Pharmaceuticals in Australia through the 2000s as an obesity therapy. The pivotal Phase 2b trial enrolled approximately 536 subjects over 24 weeks across multiple doses. It failed. AOD-9604 did not produce statistically significant weight loss versus placebo at any of the doses tested, and Metabolic Pharmaceuticals discontinued obesity development around 2007. The compound has a Phase 1 safety paper, a 2004 development-program review by Wilding documenting the trials as they were underway ², and a subsequent pivot toward intra-articular osteoarthritis applications — including a 2015 rabbit model paper showing intra-articular AOD-9604 (often combined with hyaluronic acid) reduced cartilage degeneration scores ¹⁰.

That OA work is interesting. It is also irrelevant to the case for stacking AOD with a GLP-1. Intra-articular injection in rabbits is not the same evidence base as systemic subcutaneous injection in humans for fat loss.

Fragment 176-191’s human evidence base is essentially zero controlled trials. Mouse and cell work supports lipolytic activity. Clinic and community use is real, observation is real, and the lever is real-feeling to people on it — but the formal trial record at the standard we’d apply to the GLP-1s simply isn’t there.

That’s the editorial honest tier: the only adequately powered human obesity trial of either compound was negative, the company that ran it stopped developing the molecule for obesity nearly two decades ago, and the post-mortem on what changed since then in the evidence base is, frankly, not much.

The stack itself: what we don’t have

A direct literature search returns zero published randomized controlled trials of AOD-9604 or Fragment 176-191 combined with semaglutide, tirzepatide, or retatrutide. ClinicalTrials.gov has no registered combination trials at the time of this writing. Clinic case reports exist; social-media testimony exists; vendor marketing copy exists. None of those rise to the level of evidence the GLP-1 trials themselves rely on.

Anyone telling you the combination is “proven” to preserve lean mass or reverse facial volume loss is overstating the record. The honest framing is mechanism-plausible, evidence-absent.

The female axis — what shifts

Female metabolism doesn’t make this stack work better or worse mechanistically. It shifts which question is binding.

Estrogen broadly favors subcutaneous fat storage in the gluteofemoral depot and modulates lipoprotein lipase activity across the cycle. Lipolytic responsiveness to catecholamine signaling tends to be stronger in the follicular phase, weaker in the luteal phase when progesterone-mediated fluid retention and shifted metabolic substrate use mask body composition changes on bioimpedance. None of this gives AOD/Fragment a special role. It just makes phase-aware tracking matter.

Women on aggressive GLP-1 protocols often report disproportionate loss from the face and breast tissue while gluteofemoral fat is comparatively preserved — consistent with the regional adipocyte-hormonal sensitivity pattern. There is no published evidence AOD/Fragment biases lipolysis to spare any particular depot or restore any particular depot. The lipolytic signal, if it exists in humans, is global.

Post-menopause changes the cost-benefit. Estrogen withdrawal shifts fat distribution toward visceral and increases sarcopenia risk. In that physiology, lean preservation is the binding constraint, not additional lipolysis. The marginal value of any lipolytic adjunct goes down. The marginal value of resistance training and protein goes up.

Lactation is an absolute exclusion across the GH-derived peptide category. The Manual maps the per-phase tracking and the post-menopause adjustments in detail; the article-level point is the framework.

What actually preserves lean mass

This is the unsexy section, and the section the cardiology and clinical-nutrition literature is unambiguous about.

The Locatelli 2024 Diabetes Care review on incretin therapy and body composition recommends supervised resistance training as the default adjunct, citing evidence that programs longer than ten weeks elicit roughly three kilograms of lean gain and twenty-five percent strength improvement ⁶. The protein literature broadly supports targets in the range of 1.6 to 2.2 grams per kilogram of reference body weight for active adults in caloric deficit. These two levers, run consistently, are the only interventions for which strong human evidence exists to preserve lean mass during GLP-1-induced weight loss.

That isn’t the answer anyone wants when they’re already shooting a weekly drug and looking at their face change. It is, however, the evidence-based answer. Most articles on this topic skip past this section because it doesn’t sell. We won’t. If you’re not hitting protein and not lifting twice a week, you don’t have an adjunct question. You have a foundation question.

The competitor quietly building

Worth flagging because most peptide-focused articles ignore it.

Bimagrumab is a fully human monoclonal antibody that blocks activin type II receptors, which downstream blocks myostatin signaling. Clinical trials have shown increases in lean mass and reductions in fat mass; it is now being studied explicitly as a lean-preserving adjunct to GLP-1 therapy and is discussed in the 2026 Lempesis & Dalamaga obesity-pharmacotherapy review as a credible next-generation modulator ⁹.

That doesn’t make bimagrumab safe, available, or right for anyone. It does make the pharmacological pipeline a real one with Phase 2 data and a regulatory pathway, which is more than AOD-9604 can claim for the same use case. If the “lean-preserving GLP-1 adjunct” category is going to be solved by a drug at all, the smart money in 2026 is not on AOD-9604.

Vendors, mislabels, WADA

Three things worth being explicit about.

The labeling is a mess. “AOD-9604” and “HGH Fragment 176-191” are sold by overlapping vendors. Sometimes they are the same compound under different labels. Sometimes they are genuinely different molecules (with vs. without the N-terminal tyrosine). Sometimes they are mislabeled in either direction. Assay-verified product is the exception, not the rule. Underdosing is endemic across the gray-market peptide category; a vial labeled at five milligrams may contain two and the rest filler.

Compounding regulation has tightened. The September 2024 FDA decisions and the 2024 Pharmacy Compounding Advisory Committee recommendations have left the U.S. 503A compounding picture unsettled for the GH-derived fragment class. The research-chemical channel fills the gap. A clean Certificate of Analysis should include HPLC purity, mass spec, an endotoxin result, and a third-party lab signature rather than a vendor-internal one. The verification framework belongs in The Manual, not in an article that’s trying to be brief.

WADA. AOD-9604 sits under Section S0 of the WADA Prohibited List — the “non-approved substances” category — because no government health authority has approved it for human therapeutic use. Prohibited at all times, in and out of competition ¹¹. For any reader who competes in a tested sport, this stack is non-negotiable. Not as a value judgment; as a fact.

If you’re going to do it anyway

Most readers who get this far have already decided. That’s the honest read of how this category works. The editorial line is to make sure they monitor the question rather than guess at it.

At the article level, the framework is straightforward. Foundation first — protein and resistance training, verified hit. GLP-1 ramp-up runs under clinician supervision. Mid-course body composition tracked with DEXA, not just the bathroom scale and a phone selfie. If lean loss is tracking worse than the SURMOUNT-1 75:25 ratio despite the foundation, the adjunct conversation becomes legitimate. The moment that makes the most mechanistic sense for layering AOD/Fragment in is the post-GLP-1 taper window, where the deficit is stabilizing and substrate-bias signals have something to land on.

DEXA bookends. Grip strength dynamometry monthly. Standardized facial photography. IGF-1 panel at baseline and after introducing AOD/Fragment — flat IGF-1 is the entire value proposition of the molecule; if it moves, the product is contaminated or mislabeled. The Manual maps the sequencing in detail; the article-level position is don’t add an adjunct you can’t measure.

References

1. Heffernan M et al. The effects of human growth hormone and its lipolytic fragment (AOD9604) on lipid metabolism following chronic treatment in obese mice and β3-AR knock-out mice. Endocrinology. 2001. DOI.
2. Wilding JPH. AOD-9604 (Metabolic). Curr Opin Investig Drugs. 2004. PubMed.
3. Wilding JPH et al. Once-Weekly Semaglutide in Adults with Overweight or Obesity (STEP 1). N Engl J Med. 2021. DOI.
4. Impact of Semaglutide on Body Composition in Adults With Overweight or Obesity: Exploratory Analysis of STEP 1. PMC.
5. Look M et al. Body composition changes during weight reduction with tirzepatide in the SURMOUNT-1 study. Diabetes Obes Metab. 2025. DOI.
6. Locatelli JC et al. Incretin-Based Weight Loss Pharmacotherapy: Can Resistance Exercise Optimize Changes in Body Composition? Diabetes Care. 2024. DOI.
7. Humphrey CD, Lawrence AC. Implications of Ozempic and Other Semaglutide Medications for Facial Plastic Surgeons. Facial Plast Surg. 2023. DOI.
8. Haykal D et al. The Role of GLP-1 Agonists in Esthetic Medicine: Exploring the Impact of Semaglutide on Body Contouring and Skin Health. J Cosmet Dermatol. 2024. DOI.
9. Lempesis IG, Dalamaga M. Obesity pharmacotherapy reimagined: multi-receptor agonists and next-generation metabolic modulators. Metabol Open. 2026. DOI.
10. Kwon DR, Park GY. Effect of Intra-articular Injection of AOD9604 with or without Hyaluronic Acid in Rabbit Osteoarthritis Model. Ann Clin Lab Sci. 2015. PubMed.
11. World Anti-Doping Agency. WADA Statement on AOD-9604. WADA.
12. Melson E et al. What is the pipeline for future medications for obesity? Int J Obes. 2024. DOI.
13. Pujol Calafat A et al. Semaglutide vs liraglutide real-world body composition. Nutr Hosp. 2024. DOI.
14. Blanco Anesto J et al. Tirzepatide in real-world clinical practice: body composition and muscle function. Nutr Hosp. 2026. DOI.