Ozempic and Mounjaro weren't invented for weight loss: the proof is in three trials and one bite

La domanda sbagliata al banco della farmacia

Anyone walking into a pharmacy with a prescription in hand almost always asks the same thing: is this the shot that makes you lose weight? It is the natural question. And it is the one that leads you astray.

Weight, says Amin Hedayat, is the least interesting part of the whole affair. The real question has nothing to do with pounds. It concerns a signal — a message the body sends itself every day, one that here gets turned up until it registers across the entire system: in the heart, the kidneys, the stomach, the night.

It is the claim the American doctor presses hardest, even as medicine keeps prescribing the drugs under their old name. Calling them "weight-loss shots" is an oversimplification, he argues; what is really being amplified is a hormonal signal. Hedayat also sells his own paid programme, the "Get Uninflamed Inner Circle."

Then come the numbers. Three clinical trials of very different size, different populations, different aims — and that is where you see what these drugs actually do to the body of the person taking them.

But first the story has to go back. Far back, to a bite. Because the molecule that started this entire family did not come out of anyone's imagination: it was drawn, on a laboratory bench, from the venom of a desert lizard.

The venom that spoke the language of the pancreas

The weekly injection has a date of birth, and it is not recent. In 1986, at Massachusetts General Hospital, Joel Habener, working with Svetlana Mojsov of Rockefeller University, brought into focus the active form of GLP-1: a fragment that splits off from the glucagon precursor, like a sentence cut out of a longer speech.

A hormone that has been identified, however, remains a hypothesis until someone manages to weigh it. Copenhagen took care of that. Jens Juul Holst and his team developed the radioimmunoassay that made the hormone measurable at last, and with it they demonstrated that the intestine really does secrete it. From that moment GLP-1 stopped being an idea and became a quantity.

Then comes the bite. In the archives of the National Institutes of Health a curious observation had lain sleeping for years: the venom of certain snakes and lizards enlarged the pancreas of laboratory animals. A marginal detail, the kind that stays at the bottom of a drawer. But if the pancreas reacted, then something in that venom spoke its language.

In 1992 John Eng, a physician at the Veterans Affairs, went looking for that something together with Jean-Pierre Raufman. And found it: a molecule he named exendin-4. It does what human GLP-1 does — it pushes the pancreas to produce insulin — with one difference that changes everything in practice: it lasts far longer in the blood.

The reason lies in the length. Thirty-nine units against thirty-one: the venom's chain carries an extra tail that protrudes past the end of the human one. It is a close relative, similar enough to fit the same lock. And that surplus tail is what explains why it endures so long in the blood.

From there to the vial was still a long road. Exenatide, the synthetic version of the venom's molecule, was approved in 2005.

Seventeen thousand hearts, and then the kidneys

A drug is worth as much as the people it leaves better: how many, over what time, against those who received a placebo. The two large trials in this class counted different things, in different populations.

SELECT enrolled 17,604 people with a body mass index of 27 or higher and cardiovascular disease already present, without diabetes. Treatment lasted on average 33 months. Here the drug was put to the test on hearts rather than on scales.

The target was more than a single mark. It was three events added together — death from cardiovascular causes, non-fatal heart attack, non-fatal stroke — because each one, counted alone, is too rare to say anything solid. Those who received semaglutide met that target in 6.5% of cases, against 8.0% for those on placebo. In the treated group, events arrived at 80% of the pace of the other group: hazard ratio 0.80, with a margin of uncertainty running from 0.72 to 0.90.

FLOW looked elsewhere: the kidneys, in people with type 2 diabetes and chronic kidney disease. The primary outcome arrived in 331 patients in the semaglutide arm against 410 on placebo — 5.8 cases against 7.5, per hundred years of observation pooled together. Hazard ratio 0.76, from 0.66 to 0.88. A 24% reduction.

In both cases the whole margin of uncertainty stays below one: the measured effect stands apart from chance. And the convergence is the striking part — two different organs, two different populations, one without diabetes and the other with, and the same direction.

Then comes the flip side, and it sits inside the same trial. In SELECT, cardiovascular death taken on its own does not hold: hazard ratio 0.85, with a margin running from 0.71 to 1.01. That 1.01 at the top is everything. The upper end crosses one, and that figure, alone, cannot stand.

Within the same trial, though, two other pieces hold very well indeed: heart failure, 0.82, from 0.71 to 0.96; death from any cause, 0.81, from 0.71 to 0.93. And in FLOW, mortality from any cause comes out 20% lower. A target made of several events added together is a sum of pieces that hold unevenly, and which ones hold changes everything.

The promise that comes from sleep

It travels well on social media, remarkably well: up to half of people stop choking in their sleep. It's the most striking number in the whole story. It's also the one that demands the closest look, because everything hinges on two words: "resolution" and "up to."

The figure comes from the two SURMOUNT-OSA trials on tirzepatide, presented in June 2024. At the highest dose, 43.0% of participants in Study 1 and 51.5% in Study 2 met the criteria for disease resolution. In the placebo arms: 14.9% and 13.6%.

But "resolution" here isn't a single thing. It's two doors, and passing through either one counts. The first: dropping below five breathing interruptions per hour of sleep, measured by the apnoea-hypopnoea index. The second: staying between 5 and 14, but reaching evening without the drowsiness that closes your eyes at a desk, behind the wheel, in front of the television — a score of 10 or less on the Epworth scale.

Then there's the difference between the two studies, and that's the part that lights everything up. Study 1 enrolled people who didn't use the positive-pressure machine — the one that keeps the airways open at night. They reached 43.0%. Study 2 enrolled people who kept using it every night. They reached 51.5%.

The bigger number, the one that circulates, comes from there: from people who added the drug to a therapy they were already doing. "Up to" is an honest phrase, as long as the reader knows what it marks: the maximum, not the average. Anyone opening the box expecting 51.5% is looking at the result of two things working together, the injection and the mask.

Twenty per cent of what

The headline says: twenty per cent fewer major cardiac events. That is true. But that twenty never counts the people who go home safe: it measures the distance between two percentages. In SELECT the event showed up in 8.0% of those on placebo and in 6.5% of those on semaglutide. Between the two columns sit 1.5 points per 100 people over 33 months.

Two frames for the selfsame evidence: «−20%», or «6.5 against 8.0». Neither lies. They produce two different readers. There is a still more concrete way to put it: out of 100 people with those characteristics, treated for 33 months, roughly one and a half avoid a heart attack, a stroke, a cardiovascular death. Turned inside out, the sentence runs like this: you have to treat roughly 67 people for almost three years for a single one of them to escape the event. One hundred divided by one and a half. End of the calculation.

Sixty-seven is no disappointing number. For a population at high cardiovascular risk, and for a benefit of that weight, it is a ratio cardiologists know well and consider meaningful. Yet it is a different number from the one that stays in your pocket after reading «twenty per cent», and in that gap — between two formulations, both correct — marketing lives.

On FLOW the arithmetic repeats itself. The minus twenty-four per cent, translated into absolute figures, comes to 1.7 fewer events per hundred people followed for a year. The relative is large, the absolute is measured. They are the same thing said twice, and only one of the two says how many people.

From one bite to the whole body

The chain begins at the table. You eat; the intestine releases GLP-1; the hormone reaches the pancreas and calls for insulin, slows the emptying of the stomach, lights up the brain's satiety centres. This is ordinary physiology: it happens after every meal, in every body, and it always has.

The drugs in this class invent nothing. They take that signal and amplify it, keep it burning longer. That is why the effects spring up wherever the signal travels, well beyond what any scale can show.

There is one branch where the ground narrows: the menstrual cycle and PCOS. Hedayat is precise about it: in women with PCOS something in their cycles and hormones does move, but these are observations gathered on small groups, and they need more careful handling than the findings on weight and insulin.

Then comes the kidney chapter. What FLOW counted has precise boundaries: it includes a fall in the glomerular filtration rate of at least 50% from where it started, and the only deaths that enter the count are those from renal or cardiovascular causes.

And there are the boundaries of use, which belong here, in the middle of the story, instead of at the bottom of the page. Anyone with a personal or family history of medullary thyroid carcinoma — a rare tumour, one that arises from particular cells of the gland — or of multiple endocrine neoplasia type 2, an inherited condition that predisposes to tumours of several endocrine glands at once, does not take these drugs. Full stop.

It is an absolute contraindication, printed in a black box: the most severe warning the FDA uses. And it does not concern one molecule alone. It holds for semaglutide, dulaglutide, liraglutide, tirzepatide and extended-release exenatide.

Il credito arrivato con trentotto anni di ritardo

Nel settembre del 2023 esce il ritratto di una ricercatrice che si batteva perché il proprio ruolo nell'identificazione del GLP-1 venisse riconosciuto. Un anno dopo, nel 2024, il Lasker~DeBakey Clinical Medical Research Award andava a Svetlana Mojsov insieme a Habener e a Lotte Bjerre Knudsen. Il nome scritto per esteso, sul podio, nella sua forma giusta.

Un premio, tre nomi, tre mestieri. Sullo stesso podio salgono un medico ospedaliero, una chimica della Rockefeller e una ricercatrice di Novo Nordisk, l'azienda. È una fotografia rara: accademia e industria premiate insieme per la stessa catena di lavoro, senza che nessuna delle due parti debba fingere di aver fatto tutto da sola. Il camice del reparto, il banco del laboratorio universitario, la scrivania di chi lavora dove i farmaci diventano prodotti: tre strade diverse che portano allo stesso posto.

E poi c'è la lucertola, che sul podio non sale. Il nome della molecola dice il posto da cui viene: exendin-4 vuol dire, alla lettera, il quarto composto della serie «exendin». Al Natural History Museum di Londra la chiamano oggi «il mostro il cui morso salva vite». Un animale che passa oltre il novanta per cento della propria esistenza sottoterra, al buio, dentro il fresco del deserto — e che ha finito per prestare la sua chimica a milioni di ricette.

The last stretch of the bridge, in the dark

We return to the pharmacy where this began, to the prescription clutched in the hand and to the question that comes with it. The answer is neither yes nor no. It is that this is an amplified hormone, and that what was counted has precise borders: a heart, a kidney, a night's sleep.

The drug as a bridge towards habits that last: that is an image, and as an image it should be held. The first stretch of the bridge, though, stands in plain view. The 33 months of SELECT. The months of FLOW. The 52 weeks of Study 2 on apnoea. Built spans, piers counted one by one, beneath the feet of the people walking over them.

Then the light runs out. The bridge continues past the last lamp post, and whoever comes off the drug walks a stretch where the measuring tape has never arrived.

Thirty-four years ago, in a Veterans Affairs laboratory, a vial of lizard venom was a curiosity. Today its descendant is injected once a week in millions of homes: a brief gesture, a Thursday evening or a Sunday morning, and done. What it did in 17,604 hearts over 33 months, we know with precision. What it does across an entire lifetime, the Gila monster has not said yet.