CROSS-COMPOUND BRIEFING
Three Routes Into the Satiety Loop, One of Which Bypasses It
Where each compound enters the appetite pathway, what grade of evidence stands behind it, and why the differences in mechanism track the differences in outcome.
The short version
Three peptides, three different entry points into the body's fullness system — and the results line up with the entry points.
Semaglutide — the active ingredient in Ozempic, Wegovy and Rybelsus — switches on one gut-hormone receptor, the GLP-1 receptor. That slows the stomach and reaches the brain regions that end meals. Large trials show substantial weight loss and improvements in heart and kidney outcomes [4][3][2].
Retatrutide switches on three receptors at once. Two of them are appetite receptors doing much the same job; the third, the glucagon receptor, raises the energy the body spends rather than lowering what it takes in. In Phase 2 it produced the largest weight change of the three, and also raised heart rate [11].
AOD-9604 does not touch appetite at all. It was aimed at fat cells directly. It behaved well in mice [16][17] and was well tolerated in people [15], but its human obesity programme was discontinued after the pivotal trial did not show significant weight loss.
The simplest reading: in human trials, the compounds that reach the appetite circuits are the ones that changed body weight.
Side by side
| Dimension | Semaglutide | Retatrutide | AOD-9604 |
|---|---|---|---|
| Receptor targets | GLP-1 receptor (single incretin arm) | GIP, GLP-1 and glucagon receptors (triple agonist) | No incretin receptor; acts on adipose tissue, does not bind the growth hormone receptor |
| Entry point in the satiety loop | Gut, stomach, brainstem and hypothalamus [6] | Same appetite arms, plus a glucagon arm outside the loop [8] | Outside the loop entirely — adipocyte lipogenesis and beta-3 adrenergic signalling [16] |
| Peak reported weight change | -14.9% at 68 weeks vs -2.4% placebo [4] | -24.2% at 48 weeks vs -2.1% placebo [11] | No significant human weight loss demonstrated |
| Evidence maturity | Multiple large randomised outcome trials [3][2][7] | Phase 2 complete, Phase 3 ongoing [8] | Rodent efficacy [16][17]; human trials were safety-focused [15] |
| Also studied for | Type 2 diabetes, cardiovascular outcomes, chronic kidney disease, metabolic liver disease | Type 2 diabetes [12], metabolic liver disease [10] | Preclinical cartilage repair in a rabbit model [13] |
| Dominant adverse pattern | Gastrointestinal, concentrated at escalation; nausea in about one-third [5] | Gastrointestinal, dose-related; nausea up to 45% at the top dose, plus dose-dependent heart-rate rise [11] | Tolerability indistinguishable from placebo [15] |
| Regulatory status | Approved across several indications | Investigational; not approved anywhere | Not approved anywhere; prohibited in sport at all times |
| Brand names | Ozempic, Wegovy, Rybelsus | None — investigational, identified as LY3437943 | None — not approved for any indication |
Where each one acts
The satiety loop runs gut to stomach to brainstem to hypothalamus, with adipose tissue sitting outside it as a downstream store rather than a signalling node.
Semaglutide occupies that loop at every station simultaneously. Its glycaemic effects come from the pancreas, its early fullness from delayed gastric emptying, and its weight effect predominantly from central access — the arcuate nucleus, area postrema and parabrachial nucleus, where food intake fell and food preference shifted without energy expenditure falling [6].
Retatrutide keeps that entire arrangement and adds a second kind of mechanism. Its glucagon arm increases energy expenditure and mobilises lipid, which is a metabolic-rate intervention rather than an appetite intervention [8]. Its receptor pharmacology underlines the point: it is roughly 8.9 times more potent than native GIP at the GIP receptor but only 0.4 times as potent as GLP-1 at the GLP-1 receptor [9]. It outperforms single-arm agonism while being a weaker agonist at the classic satiety receptor.
AOD-9604 never enters the loop. Its documented actions are inhibition of fat synthesis and up-regulation of beta-3 adrenergic receptor expression in white adipose tissue, with the chronic weight effect in mice abolished when beta-3 signalling was knocked out [16]. It is a fat-cell intervention with no appetite component at all.
Evidence maturity is not equal
The three occupy different tiers of knowledge, and comparing their headline percentages without saying so would be the most misleading thing this page could do.
Semaglutide has completed large randomised outcome trials — 17,604 participants for cardiovascular outcomes [3], 3,533 for kidney outcomes [2], 3,297 in a diabetes cardiovascular trial [7], 1,961 for weight management [4]. Those are event-driven trials measuring things that matter clinically, not surrogate endpoints.
Retatrutide's largest published weight trial enrolled 338 participants over 48 weeks [11]. It is a well-conducted Phase 2 study, and Phase 2 is where obesity pharmacology has repeatedly looked more decisive than it turned out to be. The cardiovascular and kidney outcome trials are running and have not reported.
AOD-9604's human dataset is roughly 900 participants across about six trials [15] — larger than retatrutide's published weight trial, and pointed at safety rather than at efficacy. A large safety dataset attached to a failed efficacy programme is not the same thing as a large efficacy dataset, and the difference is exactly what separates it from the other two.
A useful discipline when reading any of the three: ask what the trial was powered to detect, not just what it reported.
Why satiety fades, and what that implies
The satiety framing explains something the raw percentages do not: what happens when the compound stops.
Because the incretin agonists act by sustaining a signal rather than by resetting a set point, the effect is contingent on continued signalling. Semaglutide's weight effect runs through reduced intake rather than raised expenditure [6], and discontinuation is followed by substantial weight regain with cardiometabolic improvements reverting toward baseline. That is why the literature frames this class as chronic rather than curative therapy. For retatrutide there is no published discontinuation data at all — the question is open, not answered.
A second implication concerns tolerability. The signal that terminates a meal and the signal that produces nausea run through overlapping brainstem circuitry, which is why the class's commonest adverse effect is not incidental to the mechanism but an expression of it [5]. Escalation schedules exist because of this.
A third concerns what is lost. Body-composition work across the class reports that part of the weight lost is lean mass, in both semaglutide and retatrutide programmes, which is why protein intake and resistance training appear in the research conversation as a preservation question rather than as lifestyle advice.
None of these follows from AOD-9604's mechanism, because none of them applies to a compound that never engaged the circuit.
What none of them settles
Several questions remain open across all three, and this page would be dishonest without listing them.
Durability after discontinuation is unresolved for retatrutide and unfavourable for semaglutide. Long-term cardiovascular and renal safety for a triple agonist is genuinely unknown pending the ongoing outcome trials [8]. The lean-mass question has been observed but not yet answered with a protocol. Whether an adipose-tissue mechanism like AOD-9604's could work in humans given a different route, a longer half-life, or a different design has not been tested — the programme was discontinued rather than optimised [14].
And the largest question is the one the comparison itself raises: whether the appetite circuit is the only reliable lever, or merely the first one that industry learned to pull.