Two minutes to seven days

When an intestinal L-cell releases GLP-1 after a meal, the molecule has roughly two minutes to do its job. Then it is gone — not diluted, not slowly excreted, but actively taken apart. Compare that to semaglutide, which carries the same message to the same receptor and lingers for about a week. Same signal, same basic shape, five thousand times the staying power.

It is tempting to read that gap as chemistry brute-forced — a sturdier molecule, harder to break. It isn’t. Nearly every long-acting peptide on the market is still perfectly destructible; it simply spends most of its life attached to something the body has already committed to protecting. Understanding that inversion explains the entire modern injectable class, why the cadence keeps stretching from daily to weekly to monthly, and why these molecules are far harder to manufacture than their sequences suggest.

A short half-life is not a design flaw. A hormone is a sentence, and a sentence that never ends stops carrying information — the body needs signals it can switch off as fast as it switches them on. So it runs two disposal systems in parallel, and any peptide drug has to survive both.

• Proteases cut it. the enzyme DPP-4 patrols the bloodstream and the capillary walls of the gut, and it snips two amino acids off the front end of anything presenting the right pattern. GLP-1 presents exactly that pattern. The clipped product still exists — it just no longer fits its receptor, which is a more elegant off switch than destruction
• The kidney filters it out. the glomerulus works like a sieve with a size cutoff in the tens of kilodaltons. GLP-1 weighs about 3.3 kDa. It passes straight through into urine, essentially unopposed. Even a completely protease-proof version of the hormone would still vanish within hours on this route alone

That pairing is why early attempts at long-acting peptides kept disappointing. Defeat the enzyme and the kidney takes what’s left; defeat the kidney and the enzyme gets there first. Each trick below buys roughly an order of magnitude, and the weekly drugs are what you get when you stack them.

DPP-4 is precise about what it cuts, which means it can be fooled by a single substitution near the N-terminus. Nature demonstrated this before any pharmacologist did. Exenatide is synthetic exendin-4, a peptide from Gila monster venom that happens to activate the human GLP-1 receptor while carrying a different amino acid at the position DPP-4 checks. The lizard peptide is not a cleverer drug than GLP-1 — it just isn’t legible to the enzyme, and that alone stretches two minutes into a couple of hours.

Semaglutide does the same thing deliberately, swapping in an unnatural amino acid at that position so the protease finds nothing to grip. Useful, and necessary — but on its own it is worth hours, not days. The kidney is still waiting.

The second trick is the one that changed the field. Attach a fatty acid chain to the peptide, and the fatty acid does what fatty acids do in blood: it binds albumin, the most abundant protein in plasma. The peptide is now, most of the time, a passenger on a 66 kDa carrier — far too large for the kidney’s sieve, and largely shielded from circulating enzymes while docked.

Crucially the binding is reversible. At any moment a small fraction rides free, active and available to the receptor, while the bulk sits in reserve. Albumin becomes a circulating depot that meters the drug out continuously — the reason a weekly injection produces a fairly steady exposure rather than a spike followed by nothing.

The evolution of that chain is the story of the class in miniature. Liraglutide carries a 16-carbon chain on a short spacer and reaches about thirteen hours: once daily. Semaglutide lengthens the chain to an 18-carbon diacid and adds a longer, more flexible spacer between peptide and fat, which tightens albumin binding considerably — about a week. Tirzepatide pushes to twenty carbons. Three drugs, one idea, refined by a few atoms at a time.

Here is the part that turns a collection of tricks into a single principle. Albumin is not merely large — it is rescued. Cells constantly sample plasma proteins into internal compartments bound for degradation, and a receptor called FcRn reaches into those compartments, grabs albumin and antibodies specifically, and carries them back out to the bloodstream before they can be broken down. It is a salvage system, and it is the reason albumin and IgG survive for around three weeks while a naked peptide of the same journey lasts minutes.

So a fatty-acid chain is not really armor. It is a boarding pass. The peptide doesn’t out-engineer clearance; it attaches itself to a molecule the body has already decided is worth recycling, and inherits that decision.

Seen that way, the newer scaffolds are the same move with a bigger vehicle. Fuse a GLP-1 analog directly to an antibody fragment and you skip the middleman, boarding the recycling system yourself — that is how dulaglutide reaches several days without any fatty acid at all. Push further and you get maridebart cafraglutide, a full antibody carrying GLP-1 peptides as cargo, with a half-life measured in weeks and a dosing cadence measured in months. The same logic scaled up: the drug lasts as long as the thing it is riding.

There is a third lever that has nothing to do with the molecule at all. Instead of slowing how fast a drug is cleared, slow how fast it arrives. Deposit it under the skin in a form that dissolves grudgingly — microspheres, a gel, an implant — and the reservoir, not the kidney, sets the duration.

Leuprolide is the classic case: the peptide itself clears in hours, yet depot formulations cover one to six months, because the polymer matrix releases it a fragment at a time. Lanreotide uses a self-assembling gel to the same end. The peptide was never made durable — only slow to leave the injection site. The fatty-acid drugs quietly borrow this too: they aggregate at the injection site and dissolve gradually, so absorption and albumin binding stretch the curve together.

This produces a genuinely confusing consequence. When absorption is slower than elimination, the half-life you observe is really the release rate wearing elimination’s clothes — the number describes the depot, not the drug. It is worth knowing before comparing two half-lives as though they measure the same thing. Our half-life calculator makes the downstream consequences visible: how many doses until concentrations plateau, how high the accumulation runs at steady state, and how far the peak-to-trough swing travels between injections.

Stretching a half-life is not simply a convenience upgrade. It changes the shape of the exposure curve, and with it the character of the drug. Short half-lives mean tall peaks and deep troughs; a long one flattens the ride, which for this class matters directly, since the peaks are where much of the nausea lives and the troughs are where the effect fades. Smoother is not just gentler — it is what allows the effective range to be reached at all.

It also explains a cost structure that surprises people. A fatty acid on a precise side chain, joined by a purpose-built spacer, is not a step you append to a synthesis — it is extra chemistry followed by extra purification, on a molecule whose value depends entirely on what survives the process. The sequence of semaglutide is public. The reason it is hard to make well has never been the sequence.

And it sets up the next frontier cleanly. Once you accept that duration comes from the vehicle rather than the peptide, the roadmap writes itself: bigger vehicles for monthly cadence, and vehicles rugged enough to survive a different route entirely — which is the thread running through the oral pill and through every sideways move this field has made instead of counting upward. The receptor was solved years ago. Almost everything since has been a delivery problem wearing a pharmacology costume.