For fifty years, glucagon was the hormone you wanted less of. Since Roger Unger's bihormonal hypothesis in 1975, essentially every diabetes drug has treated it as the thing to suppress. The most potent obesity molecule in late-stage development contains an agonist for it.
Here is the thesis. Glucagon receptor agonism is not a third appetite lever bolted onto GLP-1 and GIP. It is the first component in this entire drug class that operates on the output side of the energy equation rather than the intake side. Semaglutide, tirzepatide, orforglipron, the amylin combinations — all of them reduce how much food gets eaten. Glucagon changes what the liver does with the substrate already in the body.
I think that distinction explains why the field cannot simply turn glucagon up, and why the real engineering problem in retatrutide is not which receptors it hits but in what proportion. The glucagon receptor has no independent therapeutic window in these molecules. Its window is manufactured for it by the GLP-1 receptor sitting alongside. The ratio, not the receptor count, is the invention.
The minimum background. Glucagon and GLP-1 come from the same gene. Proglucagon is cut differently depending on the tissue: prohormone convertase 2 in pancreatic alpha cells yields glucagon, while PC1/3 in intestinal L cells yields GLP-1, GLP-2 and oxyntomodulin. Nature built the co-agonist first. Oxyntomodulin is a naturally occurring GLP-1/glucagon dual agonist that has been circulating in human blood after meals for as long as there have been humans. The pharmaceutical industry spent two decades reverse-engineering something the gut already does.
The proof of concept was a physiology experiment, not a screening hit
In 2013, Tan and colleagues at Imperial College published a co-infusion study in Diabetes that, in retrospect, is the founding document of this entire drug class. They infused glucagon into healthy volunteers and measured what happened. Energy expenditure went up. So did blood glucose, which is exactly the problem that had kept glucagon on the villain list since Unger.
Then they infused GLP-1 alongside it. The energy expenditure gain survived. The hyperglycemia did not.
That single result is the pharmacological premise underneath survodutide, pemvidutide, mazdutide, cotadutide and the glucagon arm of retatrutide. It was demonstrated in humans with an intravenous line and a calorimeter more than a decade before any of those molecules reached Phase 3. Bloom's group had already shown the direction eight years earlier: a 2005 oxyntomodulin trial produced modest weight reduction over four weeks with a measurable increase in activity-related energy expenditure. Small effect, right vector.
The most underrated fact about the triple agonists, I think, is that their central mechanism was not discovered by a compound library. It was worked out by human physiologists asking what glucagon does when you stop looking at it as a diabetes problem.
An output lever in a class built entirely from input levers
Glucagon raises resting energy expenditure in human infusion studies by roughly 10 to 15%. It drives hepatic lipolysis and fatty acid oxidation. It increases amino acid turnover through the liver-alpha cell axis. None of those are appetite effects, and none of them have an obvious analogue anywhere else in the incretin class.
Why that matters: every intake-side drug eventually runs into the same wall. Reduce body mass and energy expenditure falls further than the mass loss alone predicts. Adaptive thermogenesis is the reason weight-loss curves flatten, and until now the entire class fought it indirectly, by suppressing intake hard enough to stay ahead. Glucagon is the first component that pushes on the other side of the ledger.
A curve that has not plateaued at 48 weeks is a claim about mechanism, not just a claim about potency.
That is the frame I would use to read retatrutide's Phase 2 result. The 24.2% mean weight reduction at the top dose over 48 weeks, published by Jastreboff and colleagues in the New England Journal of Medicine in June 2023, is the number that got quoted. The more interesting observation is that the curve had not yet flattened when the trial ended.
The liver is where the receptor stops being subtle
The glucagon receptor is expressed most densely on hepatocytes. So the mechanism makes a testable prediction: if glucagon agonism is doing what the physiology says, molecules containing it should over-deliver on liver endpoints relative to the weight loss they produce. They do, consistently, across sponsors who have no reason to agree with each other.
Boehringer Ingelheim's survodutide, a GLP-1/glucagon dual, reported that 83% of participants at the highest dose achieved MASH improvement without worsening of fibrosis, against 18% on placebo, in its Phase 2 readout. The weight loss in that program was strong but not class-leading. The liver result was.
Altimmune's pemvidutide is built at a deliberate 1:1 GLP-1-to-glucagon ratio and its Phase 2b MASH trial reported 59.1% MASH resolution without fibrosis worsening. The sponsor markets the liver, not the scale, which tells you where they believe the value sits. Retatrutide's own hepatic sub-study reported liver fat reductions above 80% in relative terms, with most participants moving into the normal range.
Contrast that with semaglutide, which is genuinely effective in MASH but appears to get there substantially through the weight loss. The glucagon-containing molecules arrive at liver endpoints ahead of what their scale numbers would buy. That asymmetry is the strongest cross-molecule evidence available that the third receptor is contributing something the other two cannot.
There is no adding glucagon. There is only choosing a ratio
Coskun and colleagues characterized retatrutide in Cell Metabolism in 2022 as unbalanced by design: potent at the GIP receptor, with activity at the GLP-1 and glucagon receptors attenuated relative to the native hormones. That is not a compromise. That is the product.
Push glucagon too high relative to GLP-1 and you get back the problems Unger warned about: glucose drift, heart rate increases, transaminase movement. Push it too low and you have built a more expensive GLP-1 drug with a harder manufacturing story. The therapeutic window is a construction, and it collapses in both directions.
Cotadutide is the cautionary case. AstraZeneca carried a GLP-1/glucagon dual through Phase 2b with real hepatic signals and modest weight loss, then deprioritized it. My read is that the ratio was tuned for glycemia during an era that turned out to be scored on weight — a strategy error dressed up as a pharmacology one. Mazdutide sits at the other end: a GLP-1-weighted dual that reached approval in China with weight reductions in the mid-teens. Competitive rather than class-leading, but paired with a cost structure and a market where that trade is entirely rational.
Same three-letter receptor, four different diseases being treated, depending on the dial settings.
The case that glucagon is doing less than the class implies
Now the honest counter-argument, because the evidence here is genuinely incomplete. Nobody has run the deconvolution. No trial has compared retatrutide against a matched GIP/GLP-1 molecule at equivalent exposure, which means the 24.2% could be substantially a story about how much drug patients tolerate rather than about the third receptor doing distinct work.
The costs are also not hypothetical. Dose-dependent heart rate increases show up across glucagon-containing programs. Transaminase elevations appear at higher exposures. And the liver-alpha cell axis implies increased hepatic amino acid catabolism, which sits uncomfortably beside an unresolved argument about body composition in this class.
There is a competing explanation I have made myself: that the GIP receptor's contribution is largely about tolerability, buying dosing headroom rather than adding weight loss. If that is right for GIP, it can be right for glucagon too, and part of what looks like a mechanism story is really an exposure story wearing a lab coat.
My position: the liver data is strong evidence of a distinct contribution, because liver fat responds faster and further than the weight loss explains. The weight data is not, yet. TRIUMPH is the program that settles it.
What this changes about how to read the pipeline
If glucagon is an output lever rather than a third appetite hormone, several things follow. Cardiometabolic and hepatic endpoints become the place to watch these molecules, not the scale. Trials that measure only weight are measuring the wrong axis for the component that distinguishes them. And the next generation of design questions is about proportions, not about how many receptors a peptide can reach.
Fifty years of calling glucagon the villain was not wrong about the physiology. Unger was right that unopposed glucagon is a problem. What the field got wrong was the arithmetic — the assumption that a hormone which raises glucose could only ever be subtracted. It turns out it can be added, provided you also add the thing that cancels its worst property, and provided you get the ratio right to within a fairly narrow band.
That is a genuinely new kind of drug design problem. The receptors in these molecules are not additive. They are mutually enabling, and one of them exists mainly to make another one survivable.
Ozemback — August 2026