The GLP-1 Pipeline: Every Serious Contender in One Place

A 2024 systematic review of emerging obesity drugs counted 14 ongoing phase 3 trials, and the list reads less like a drug class than like a group chat. There’s retatrutide, which hits three receptors. There’s CagriSema, which is two drugs sharing one product. There’s orforglipron, which is a pill. There are mazdutide and survodutide, which hit GLP-1 and glucagon, and ecnoglutide and TG103, which are plain GLP-1 receptor agonists, and also, sitting in the same list, dapagliflozin and a sibutramine/topiramate combination.

So here is the glp-1 pipeline in one sentence: it sorts along two questions, which receptors the molecule touches and how it gets into the body. On receptors, the options are GLP-1 alone, GLP-1 plus GIP (tirzepatide is the established example), GLP-1 plus glucagon, all three at once, GLP-1 plus amylin, and one oddball that activates GLP-1 while blocking GIP. On delivery, it’s mostly injected peptides plus a growing set of oral small molecules, led by orforglipron. Which of these wins is not settled, and a good chunk of the field is still in phase 2.

Why build another version of a drug that already works?

The sensible reaction to all this is: the drugs work already, so why are there so many of them? Tirzepatide produced mean weight reductions of 19.5% and 20.9% at its 10mg and 15mg doses in the SURMOUNT-1 phase 3 trial. GLP-1 receptor agonists reduce major cardiovascular events, and the SELECT trial extended that benefit to people with cardiovascular disease and obesity but no diabetes. At some point a product category is just done, and everybody else is making toothpaste.

That objection is reasonable, and the answer is weirder than “competition.” The pipeline is not a dozen copies of one drug. It is a dozen different arguments about what the drug should be in the first place. Some of those arguments contradict each other. Some of the contradictions are both producing weight loss in trials. That’s the fun part.

(I should admit that “reading receptor pharmacology papers on a Sunday” is how I relax, which says more about me than about the papers.)

The multi-agonists: one peptide, several doorbells

The original idea is simple. GLP-1 is a gut hormone, its receptor sits on cells in the pancreas and brain and elsewhere, and a drug that activates that receptor lowers blood sugar and appetite. Then someone noticed you could build a single peptide that rings more than one receptor. Tirzepatide is a 39-amino-acid synthetic peptide that activates both the GIP and GLP-1 receptors, a design that, according to one cardiology review, earned it the nickname “twincretin.” In phase 3 it beat GLP-1 monotherapy on blood sugar in the SURPASS trials and on weight in the SURMOUNT trials.

Once two works, three is obvious. Retatrutide is a triple GLP-1/GIP/glucagon agonist. UBT251 is another one, and in a phase 2 trial of 205 adults with obesity over 24 weeks it produced weight loss of 13.6% to 19.7%, versus 2.0% on placebo. A 2026 review said GLP-1/glucagon duals and triples like retatrutide “have exhibited unprecedented weight-loss efficacy,” which is the sort of sentence review authors get to write about once a decade.

Why glucagon, a hormone that raises blood sugar? The pitch is that GLP-1/glucagon agonism raises energy expenditure by stimulating fat oxidation, which is a different lever from the appetite-focused GLP-1/GIP approach. Mazdutide and survodutide are the duals built on that idea. The structural biology here is fiddly: cryo-EM work on dual GLP-1/glucagon peptides (including cotadutide and SAR425899) found that the first three residues of the peptide engage the two receptors differently, and the middle of the peptide reshapes a loop of the GLP-1 receptor on binding. Selectivity comes down to a few side chains pointing slightly different ways.

I’m contractually required to imagine the endpoint. Here is a made-up scenario. A medicinal chemist walks into a Monday meeting and says, “Hey guys, what if we also hit glucagon.” Everyone nods. The next Monday it’s amylin. Then calcitonin. Then, why not, the receptor for some hormone that governs how much you enjoy jazz. By year five the team has a molecule that activates every receptor in the human body, which is technically a full agonist of being alive, and the phase 1 readout is just the word “yes.” This is dumb, obviously. But the direction of travel is real.

Amylin, and the question of one molecule or two

The amylin branch makes the structural choice explicit. CagriSema is a fixed-dose combination of semaglutide and cagrilintide, an amylin analogue: two molecules sharing one product, a perfectly ordinary arrangement between consenting peptides. Amycretin goes the other way and puts GLP-1 and amylin agonism into a single molecule (one source counts calcitonin receptors as a third target). It reached 13.1% weight loss at 12 weeks given orally and 24.3% over 36 weeks given by injection.

The cleanest comparison is a restaurant bill. CagriSema is two diners asking for separate checks: each drug keeps its own pharmacology, its own dose, its own half-life. Amycretin is one check with two items on it, which is tidier to carry but means you can’t adjust one item without reprinting the whole receipt. Neither is obviously better. Which one patients and regulators end up preferring is not settled by the facts here.

The GIP paradox in the GLP-1 pipeline

Now the strange part. Tirzepatide activates the GIP receptor and adds weight loss on top of GLP-1. Maridebart cafraglutide (Amgen’s MariTide, AMG 133) is a bispecific antibody that activates the GLP-1 receptor while blocking the GIP receptor, and it is also in development because blocking GIP also seems to add weight loss. A review titled “The Premise of the Paradox” calls this “a central unresolved scientific question in incretin pharmacology,” which is the polite academic way of saying the arrows point in opposite directions and both hit the target.

One reading is that GIP biology is two-sided and both handles turn the same valve, for reasons nobody has fully written down. A second reading is that the GIP part matters less than the branding suggests, and most of the work in both drugs is being done by the GLP-1 half, with GIP along for the ride in whichever direction. The data cited here do not choose between them. I feel like that should bother people more than it does, but anyway, both programs exist and are spending money.

Pills, which are a different business entirely

The oral branch is less about biology than about manufacturing and convenience. Peptides are big and fragile and the stomach chews them up, so the alternative is a small, non-peptide molecule that fits into the receptor without being a hormone at all. Orforglipron, from Eli Lilly, is one. Lilly described it in a press release as “the first oral small-molecule GLP-1 receptor agonist to successfully complete a Phase 3 trial,” where it lowered HbA1c by 1.3% to 1.6% across doses, and the company positions it as potentially the first oral GLP-1 agonist approved specifically for obesity. Behind it are HRS-7535, which showed efficacy in a phase 2 diabetes trial in people already on metformin and SGLT2 inhibitors, and aleniglipron, with phase data published in Nature Medicine in 2026.

You and I both understand that the receptor does not care whether it was reached by a pill or a pen. The pill matters because of everything around the receptor: supply, cost, people who would prefer not to inject, or whatever else drives a prescribing decision. That’s fine. It is still a real scientific achievement to get a non-hormone to do a hormone’s job.

Bias, the drug that only delivers half the message

The last axis is the most abstract. The GLP-1 receptor signals mainly through a G protein called Gs, but it also recruits a protein called beta-arrestin, and a drug can be built to favor one route. A study of one nonpeptide agonist, LY3502970 (also called OWL833), found it binds a unique pocket in the upper part of the receptor and acts as a partial agonist biased toward G protein over beta-arrestin. Other work found that the outward displacement of one extracellular loop, ECL3, is what distinguishes G protein-biased states, and that small changes to the front end of GLP-1 itself can produce bias. One 2026-era review lists multi-agonists, biased agonists and oral formulations as the three main axes of next-generation development.

The point is that the pipeline isn’t one race. It’s three or four races run on the same track, and some runners are entered in several. Safety, so far, looks similar across the class in pooled data: a 2026 systematic review found serious adverse events in 6.5% of GLP-1 drug recipients versus 5.2% on placebo. Fine, not nothing.

And since activating GIP helps and blocking GIP also helps, the obvious next entrant is a molecule that does nothing at all at the GIP receptor, which would be the third GIP strategy, the cheapest to make, and, lol, already on the market.