Maridebart Cafraglutide: Amgen's Antibody Bet

Amgen is developing an obesity drug that blocks the GIP receptor, which is the same receptor Eli Lilly’s tirzepatide was built to switch on. Both approaches are supposed to help you lose weight when paired with a GLP-1 drug. Both do. The Amgen drug is called maridebart cafraglutide (also MariTide, also AMG 133), and a review titled “The Premise of the Paradox” describes it as “a bispecific antibody that combines GIP receptor antagonism with GLP-1 receptor agonism and is in clinical development as a novel obesity therapeutic.”

So, plainly: it is an antibody with a GLP-1-type peptide attached. The antibody end blocks GIP signaling and the peptide end activates the GLP-1 receptor. It is not approved. Because it is built on an antibody, the expected selling point is a schedule less frequent than weekly. On the specifics people search for, though (the exact dosing interval, the phase 2 weight-loss percentages), the sources behind this article give no numbers, so I will not invent any. What I can explain is the design and why it is weird.

What maridebart cafraglutide actually is

Most incretin drugs are peptides, short chains of amino acids modeled on gut hormones. Tirzepatide, for instance, is a 39-amino-acid synthetic peptide that activates both GIP and GLP-1 receptors, which got it the nickname “twincretin.” Retatrutide goes further and hits three receptors from one molecule, which took a lot of medicinal chemistry to manage. (The site has separate pieces on both, so I will leave them there.)

Maridebart cafraglutide is a different kind of object. Its core is an antibody, a large protein the immune system normally uses to grab specific targets, aimed here at the GIP receptor to hold it shut. The GLP-1 activity comes from peptides joined to that antibody. Chemists call the result a conjugate. That is a formal word for two things bound together permanently, and here, as in some marriages, the main point of the arrangement seems to be that one partner gets to stick around longer.

The dumb version: turn GIP up

Here is the simple story, the one you would tell at a dinner party (if your dinner parties are already going badly). GLP-1 is a gut hormone that lowers blood sugar and appetite. GIP is another gut hormone in the same family. If one hormone helps, two hormones should help more. So you build a molecule that activates both receptors.

That is tirzepatide, and the results were good. In the SURMOUNT-1 phase 3 trial, adults with obesity lost an average of 19.5% of body weight at the 10 mg dose and 20.9% at 15 mg. Across the SURMOUNT and SURPASS programs, tirzepatide beat GLP-1-only drugs on weight and on blood sugar. The dumb version worked.

That is about as clean as drug development gets. Pull the lever, get the result.

The sophisticated version: turn GIP down

Now the long road. Suppose you look at the GIP receptor and conclude, for reasons involving fat tissue and energy storage, that GIP is better blocked than activated. Then you would want the opposite of tirzepatide: a molecule that blocks GIP while still activating GLP-1. Activating a receptor with a peptide is the usual approach. Blocking one is something antibodies are good at. So you take an antibody that blocks the GIP receptor and attach GLP-1 agonist peptides to it.

This is the more sophisticated version. It requires a different modality, a different manufacturing process, and a theory of GIP that points the other way. And then it also produces additive weight loss when combined with GLP-1 agonism.

The same review says it directly: “The paradox that both GIPR agonism (tirzepatide) and GIPR antagonism (maridebart cafraglutide) provide additive weight loss when combined with a GLP-1 receptor agonist is a central unresolved scientific question in incretin pharmacology.”

Two companies pushed a receptor in opposite directions. They ended up in roughly the same place. It is a bit like two teams in the Super Bowl running toward opposite end zones and both being credited with the same touchdown.

Why the paradox may not really be one

I should be careful here. I said both companies pushed the receptor in opposite directions. More precisely, they designed molecules that push in opposite directions in a test tube, and what happens to the receptor inside a living body after weeks of exposure is a separate question. Biology adds time, and time can change what a signal means.

That suggests a second reading. Maybe a receptor stimulated hard and continuously eventually stops responding, so a strong agonist ends up acting a lot like an antagonist, and the two drugs converge because the cell treats them about the same. Or maybe GIP affects several tissues in different ways, and each drug happens to move the one that matters for weight, from opposite sides. The sources do not choose between these. They call it “unresolved,” and I will leave it there too.

I find this genuinely funny (in the way scientists probably do not enjoy). Picture the program meeting. “Hey team, good news, the antagonist works.” “Great, so the agonist was wrong?” “No, the agonist also works.” “So what does GIP do?” “Yes.”

Why bolt a peptide onto an antibody

The point is the calendar. Incretin drug development has been a long effort to make doses last longer, from hours to days to a week (the site has a whole piece on half-life engineering). Peptides on their own get cleared fairly fast. Antibodies are large and are generally cleared much more slowly, so attaching the peptide to one is a way of letting it stay in circulation longer. Hence the expectation of monthly rather than weekly dosing. The sources here do not give the actual schedule tested, so treat the monthly framing as the design goal, not a reported result.

To be fair, this is not the only route to convenience. A whole other branch of the pipeline is pursuing oral pills. Orforglipron, Lilly’s small-molecule pill, is the first oral small-molecule GLP-1 agonist to finish a phase 3 trial. In ATTAIN-1, a 72-week trial of 3,127 patients, it produced weight loss of 7.5%, 8.4% and 11.2% at 6, 12 and 36 mg, versus 2.1% on placebo. AstraZeneca’s elecoglipron reached up to 10.5% at week 26 in phase 2. A review of GLP-1 receptor biology says “dual and triple receptor agonists,” “biased agonist design,” and “oral formulations” are “the three major axes of next-generation incretin drug development.” An antibody you inject monthly does not fit neatly into any of them, which is a little awkward for anyone who likes tidy categories, and I write about drug pipelines, so that includes me.

One side wants a daily pill. The other wants an injection a few times a year. Between them, you could build every possible schedule.

What the phase 2 data can and cannot tell us here

This is where I would normally give you the phase 2 numbers. The sources available for this piece do not contain them. No percentages, no week counts, no dose arms. I am not going to fill that gap from memory, so whatever was reported in phase 2, this article does not settle it.

What I can give is the benchmark any number will be compared against, which is very high. Tirzepatide: about 20% in phase 3. Retatrutide: up to 28.7% at 68 weeks in phase 3. UBT251, a triple agonist from China, reached 19.7% after 24 weeks in phase 2, versus 2.0% on placebo. Novo Nordisk’s CagriSema reached 23% at 84 weeks and still failed its non-inferiority test against tirzepatide, which reached 25.5%. That last example shows how this market works. A drug can produce 23% weight loss and still be counted as a disappointment, because the comparison is to the competitor, not to placebo.

I assume, then, that the question for maridebart cafraglutide is not whether it causes weight loss. It is how it stacks up in a comparison table against drugs dosed weekly or daily. If the weight loss is comparable to the competition, the schedule becomes the selling point. If the weight loss falls short, the schedule becomes an explanation for why.

Here is a deliberately silly scenario to show how far the logic goes. Say every company keeps extending dosing intervals: weekly, then monthly, then quarterly. Eventually someone puts a once-a-year obesity antibody in a slide deck, and the main clinical obstacle is that trial participants forget which year it is. Analysts start modeling adherence on a decade basis. That is where the incentive points, anyway.

Both roads, same place

Put the two versions side by side. The simple version activates GIP and GLP-1 with one peptide and gets about 20% weight loss. The sophisticated version blocks GIP with an antibody, adds GLP-1 peptides, aims for monthly dosing, and arrives at the same basic claim: additive weight loss over GLP-1 alone. The longer road required a new modality and a contrary theory of GIP. It reached the same result, and then left the theory of GIP less clear than before.

Basically, Amgen may succeed commercially regardless of which reading of the paradox turns out to be correct, since a monthly drug sells on the schedule, and GIP biology may stay unexplained for a long time while both molecules are on the market.