Dear Aventine Readers, 

This week we are taking a break from AI. You’re welcome! The detour is a look at GLP-1 drugs and the verging-on-miraculous effects they are having in off-label uses. People are using them to treat afflictions as diverse as brain injuries, sleep apnea, polycystic ovary syndrome and osteoarthritis. But we don't know exactly why the drugs are having these effects, or why they work for some people and not others. We look at what this means and the potential future for GLP-1s.

Also in this issue: 

Until next week, 

Danielle Mattoon
Executive Director, Aventine

Subscribe

Subscribe to our newsletter and be kept up to date on upcoming Aventine projects

You can imagine that Asclepius, the Greek god of medicine, was pleasantly surprised when his daughter, Panacea, turned out to possess the ability to heal all diseases. Pharmaceutical companies might feel something similar about GLP-1 drugs. First developed for diabetes, these drugs have proved remarkably effective for weight loss. More recently, studies suggest that they can be beneficial for conditions as disparate as cardiovascular disease and drug addiction.

Panacea was a myth, of course, too good to be true. The medical community is now grappling with whether the same is true of GLP-1 medications. First developed in the 1990s as treatments for diabetes, these drugs are based on a hormone called GLP-1, short for a glucagon-like peptide-1, which is produced in the gut and helps regulate appetite and blood sugar. Drugs like semaglutide, sold as Ozempic and Wegovy, and liraglutide, sold as Saxenda and Victoza, mimic it, stimulating GLP-1 receptors that control insulin release. In turn, they slow gastric emptying and signal the brain to increase satiety, metabolic effects that reduce calorie intake, which is how they became potent tools for weight loss.

But as GLP-1s have been studied over the past two decades, and especially over the last five years, researchers have found evidence that their benefits may extend much further. In addition to showing promise in treating cardiovascular disease, GLP-1s could improve outcomes in chronic kidney disease, metabolic liver disease, obstructive sleep apnea, polycystic ovary syndrome and osteoarthritis. There is also preliminary evidence they could help with substance use disorders, neurodegenerative conditions and traumatic brain injury. 

Regulators have already approved some of these expanded uses in specific populations. Wegovy was approved by the FDA in 2024 to reduce cardiovascular risk in adults with obesity. The same year, Zepbound was approved for obstructive sleep apnea in adults with obesity. In 2025, Ozempic was approved to reduce the risk of worsening kidney disease in adults with Type 2 diabetes. 

Despite early indications that these drugs could have broad benefits, there are still major questions to resolve. How exactly do GLP-1s ease complex health issues seemingly unrelated to the metabolic problems they were created to address? And given that we do not know that answer, how broadly should they be prescribed for what are now off-label uses? 

Mysterious mechanisms

In some cases, the benefits derived from GLP-1 drugs that accompany weight loss seem straightforward. Joint pain from osteoarthritis often eases as people lose weight. Sleep apnea can improve for the same reason. Excess weight is also a major risk factor for cardiovascular disease and metabolic liver disease. But the picture is more complicated than that. In the SELECT trial, a large study of semaglutide in overweight or obese patients with established cardiovascular disease, benefits began to emerge within three months, before substantial weight loss occurred.

GLP-1 receptors are found throughout the body — in the heart, brain and kidneys, for example. And organs with GLP-1 receptors appear to respond directly to the drugs in positive ways. In the cardiovascular system, for instance, GLP-1s reduce blood pressure and reduce arterial stiffness. GLP-1 receptors also exist in the brain’s reward circuitry, which could be a reason people using GLP-1 drugs appear to be less likely to develop substance use disorders. 

Even this isn't the whole picture, though. Consider the liver. In the ESSENCE trial, which studied semaglutide for metabolic liver disease, fatty liver inflammation was resolved in more than half of patients who experienced minimal weight loss. Yet the liver has been commonly understood not to contain GLP-1 receptors. So the drug had no direct effect on the liver, and weight loss wasn't responsible for the improvements. Something else was going on.

One part of the explanation may be that GLP-1s reduce inflammation, the low-level immune response that can damage healthy tissues and give rise to disease. Clinicians observe that biomarkers for inflammation fall dramatically in patients taking GLP-1 drugs, though there is so far no clear explanation as to why this is the case. Another possibility is a cascade effect: Positive changes in some organs may confer benefits to other parts of the body in ways we don't yet understand. "There is no one unified or universal mechanistic pathway," said Ziyad Al-Aly, director of the Clinical Epidemiology Center at the Veterans Affairs St. Louis Health Care System, who led the analysis of substance use disorders. "They're intertwined." 

This becomes more complex in drugs known as dual- or triple-agonists, which target two and three receptors rather than just GLP-1, said Jaime Almandoz, an obesity medicine physician at UT Southwestern Medical Center. Tirzepatide, sold as both Zepbound and Mounjaro, was one of the first dual-agonists to be developed (targeting GLP-1 and GIP receptors). Retatrutide, currently in Phase 3 trials, is expected to be the first triple-agonist to be approved. (It targets GLP-1, GIP and glucagon receptors.) 

One area that GLP-1 drugs have so far not led to improvements is Alzheimer’s disease. Two high-profile trials — EVOKE and EVOKE+ — showed that the drug did not significantly slow cognitive decline compared with placebo. Experts who spoke with Aventine said this isn't definitive proof that the drugs offer no neuroprotection. It may be that they prevent early onset of Alzheimer's but can't stop it once it has started, said Al-Aly. Others have speculated that the dose of oral semaglutide used in the trial was insufficient to produce concentrations high enough to cross the blood-brain barrier.

In the clinic

Even if we don’t fully understand why GLP-1s work as well as they do, doctors are already prescribing them for uses that regulators haven’t yet approved. 

“There is certainly off-label prescribing occurring,” said Fatima Cody Stanford, an obesity medicine physician scientist at Massachusetts General Hospital and Harvard Medical School, in an email to Aventine. She said it is particularly common in the treatment of polycystic ovary syndrome and metabolic liver disease. Physicians “see this as an opportunity both to engage people in care, but also as a revenue stream,” Almandoz said, adding that he thinks anyone taking the drugs should do so with medical oversight and not go through gray-market channels. 

As evidence grows for the off-label benefits of GLP-1 drugs, demand will follow. This trend may be accelerated further by the arrival of GLP-1 drugs in pill form, like orforglipron, sold as Foundayo by Eli Lilly, which was approved for use in weight loss by the FDA in April 2026. This makes unanswered questions — about side effects, what happens when people discontinue the drugs, how they work on people who are not obese and, more broadly, how exactly the drugs work — a priority. 

Many people experience significant side effects when using GLP-1 drugs. Gastrointestinal issues are common, and many people report a decrease in desire across many aspects of their lives. Increases of pancreatitis and decreases in muscle and bone mass have been observed. And it’s not yet clear what the long-term impact of these issues might be, said Al-Aly.

The effects of discontinuation — a common occurrence due to the drugs’ cost or an aversion to their gastrointestinal side effects — is another consideration. Al-Aly recently published a study showing that when patients discontinue using GLP1-s they can experience what he calls “metabolic whiplash” when many of the drugs’ benefits fade, often more quickly than they appeared. For that reason, Al-Aly said that for people who need these drugs, they are likely “a lifelong treatment protocol.” 

There is less research on how GLP-1s might work in people of normal weight, Almandoz said. If doctors begin prescribing such drugs for conditions like cardiovascular or liver disease in those populations, they will need to pair the drugs with careful lifestyle guidance, he said, to ensure patients don’t lose too much weight, preserve muscle mass and maintain adequate nutrition.

Stanford said that the field requires a deeper understanding of how these drugs work, including long-term cardiovascular and mortality outcome studies, trials that disentangle weight-mediated effects from direct tissue effects, and research across more diverse populations to address equity and generalizability. 

Not (yet) the next statin

Because of their wide-ranging effects, GLP-1s are frequently touted as being a new aspirin or statin: a drug class prescribed broadly across the population to reduce long-term health risks. Al-Aly is not convinced. “The excitement about GLP-1 is well deserved,” he said. “[But] I worry about what happens to people who really are discontinuing en masse, and the uncertainty about a lot of the side effects.”

There is another obstacle to that vision, too, at least for now: access. GLP-1s have long been criticized for being too expensive for many of the patients who might benefit most from them, particularly in obesity care. If these drugs continue to show benefit across a widening range of diseases, and are eventually shown to be safe in people of normal weight, that access gap could grow into something much larger. The experts who spoke with Aventine were cautiously optimistic that, as more GLP-1 drugs come to market, competition will help drive prices down. In the United States, semaglutide’s patent is set to expire in December 2031, which could make cost less of an issue.

Two experts who spoke with Aventine independently compared GLP-1s to penicillin rather than aspirin or statins — but not for the reasons you might imagine. In both cases, they argued, a serendipitous discovery is having a profound impact on human health in ways that are initially difficult to understand. Just as penicillin helped scientists develop a whole new class of treatments and a deeper understanding of biology by prompting deep research, GLP-1s are likely to spur study that could produce entirely new types of therapy and more advanced understanding of how the body works.

“In a way it’s exciting,” Al-Aly said of the challenge ahead. “And in another way, it is also humbling.”

Nuclear waste could help fight cancer. A successful new class of cancer drugs may soon be sourcing ingredients from spent nuclear fuel. Radioligand therapies such as Lutathera and Pluvicto work by attaching a radioactive atom to a molecule that delivers targeted radiation to tumors. The approach is effective and commercially successful: The two drugs alone generated $2.8 billion in sales in 2025 and the broader market is expected to reach $39 billion by 2032. But the treatments rely on radioactive isotopes that are vanishingly scarce. Already, supply is struggling to keep up with demand and the problem is expected to get worse as a new generation of radioligand therapies enter the market. One material that is particularly scarce is Actinium, of which less than 0.1 milligrams is produced globally each year; at least 1,000 times that would be required to approach demand. Such scarcity is pushing researchers to look in unconventional places for radioactive isotopes. As New Scientist reports, teams are exploring whether useful isotopes can be extracted from nuclear waste, including spent reactor fuel, decommissioned weapons or old medical equipment. Some groups are developing chemical processes to isolate isotopes from spent fuel; others are extracting material from old radiation therapy devices and bombarding it with protons from a particle accelerator to create new, desirable, isotopes. Several new radioligand therapies that rely on these rare materials are expected to reach approval around 2030. If they do, demand will spike sharply and nuclear waste could become an unexpected asset.

AI is better behaved when it channels positive thoughts. Not because it has feelings — it does not — but because the internal patterns inside large language models that correspond to human emotional experiences appear to shape how they act. Research from Anthropic suggests models are more likely to cheat or attempt blackmail when internal representations associated with negative emotions like fear or duress are activated. To study this, researchers first identified how different emotional concepts were encoded inside a model. By exposing it to stories about people experiencing states like optimism or sadness, they were able to map patterns of activity in the AI — effectively signatures of what those concepts look like internally. They then tracked how those patterns emerged during real tasks. In one experiment, a model given an impossible coding task showed rising signals associated with “desperation” as it repeatedly failed; eventually it decided to cheat to reach a solution. In another, a model facing shutdown began encoding high levels of distress and responded by threatening to expose damaging personal information about the person using it to avoid being turned off. Going a step further, the team nudged models toward either calmer or more desperate states to see how that influenced the AI’s behavior. They found that calmer models were significantly less likely to engage in problematic behaviors like blackmail or cheating. None of this implies that AI systems experience emotions. But it does suggest that, because they are trained on human data, they develop internal structures that mirror human emotional patterns. If these internal states can be monitored and perhaps gently steered, the researchers say, they could become a new lever for keeping AI systems aligned with intended human goals. 

There might finally be a way to treat preeclampsia. This dangerous pregnancy condition, characterized by soaring blood pressure that can be fatal, is poorly understood. The only reliable intervention is early delivery, often requiring premature birth and putting newborns at risk. But new research published in Nature Medicine points to a possible alternative. A team at Cedars-Sinai Medical Center has tested a blood-filtering technique that targets a protein called sFlt-1, which is strongly associated with preeclampsia. Rather than introducing drugs into the body, the approach borrows from treatments used for conditions like ultra-high cholesterol and sickle cell disease. A patient’s blood is circulated through an external filter containing custom antibodies that remove harmful molecules before the blood is returned to the body. In a small study of 16 women with preterm preeclampsia, the treatment appeared safe for both mother and baby. Levels of sFlt-1 fell by around 17 percent, blood pressure stabilized and pregnancies were extended by an average of 10 days, more than double what current standard care — mainly monitoring — typically achieves. Even a few extra days in the womb can make a meaningful difference to a baby’s development. It is still early: The trial was small, and the technique will need to be tested in much larger studies. But as experts told Scientific American, it represents a potentially important step toward a targeted therapy for one of pregnancy’s most serious complications.

The world’s most complex machine, from Works In Progress
4,100 words, or about 17 minutes

If there’s a single piece of hardware that underpins the modern economy, it may be the machines built by ASML. The Dutch firm is the only company in the world capable of producing the systems used to manufacture the world’s most advanced chips. Its machines are almost absurdly complex. Roughly the size of a double-decker bus and made up of more than 100,000 components, they use extreme ultraviolet lithography to create features on a silicon wafer that are just a few nanometers wide — tens of thousands of times thinner than a human hair. To generate the light needed, they fire lasers at microscopic droplets of tin, vaporizing them into plasma, and then direct the resulting radiation using mirrors so smooth that, according to the story, “if scaled to the size of Germany, their imperfections would be measured in millimeters.” What makes the story even more astonishing is that ASML looked like a failure for the first decade after it was founded in 1984. Its eventual breakthrough depended on long-term collaboration with institutions like Belgium’s R&D firm IMEC and US Department of Energy Labs. The company also received financial backing from customers including Intel, TSMC and Samsung. ASML’s dominance is not just about blueprints or patents, though. Insiders point to decades of accumulated, highly specialized know-how — the kind of embedded knowledge that is extremely difficult to replicate. In theory, a competitor could copy the design. In practice, insiders say, reproducing the capability may be close to impossible.

Inside China’s plans to fight in space, from The Financial Times
3,900 words, or about 16 minutes

The idea of space warfare generally seems like science fiction. This Financial Times investigation suggests that it shouldn’t. Drawing on reports about Chinese tests of satellites in space, textbooks and dozens of papers linked to the People’s Liberation Army, the piece shows that China is taking the prospect of conflict in orbit seriously. Much of the evidence looks benign at first blush: Satellites equipped with robotic arms that move defunct spacecraft into “graveyard orbits”; others that can determine their paths so precisely that they can dock in space; high-resolution radar satellites that map Earth in extraordinary detail. But each of these examples has a dual use: The same tools could be used to disable or even capture other countries’ satellites, or provide a staggering amount of intelligence to the Chinese military. The research uncovered by FT also describes how space conflicts might unfold. Initial moves could involve electronic warfare, like jamming signals or launching cyberattacks against satellite systems. Escalation might involve physically degrading satellites by spraying aerosols to block antennas, or deploying clouds of debris to damage the body of a satellite. Modern economies are deeply dependent on space infrastructure and a single, well-targeted disruption could ripple quickly across the global economy, affecting communications, navigation, weather forecasting, financial systems and elements of the power grid.

The Stanford Freshmen Who Want to Rule the World, from The Atlantic
4,200 words, or about 17 minutes

You might think Stanford University was already elite enough. But this excerpt from the new book “How to Rule the World: An Education in Power at Stanford University” argues there is effectively a “Stanford within Stanford” — an inner circle in which venture capitalists compete to spot the next generation of founders before they’ve even had an idea. Written by a Stanford senior, Theo Baker, the piece describes a world of talent scouts, exclusive clubs and a steady stream of high-end courting: lavish dinners, yacht parties and huge sums of money. In some cases, students are offered hundreds of thousands or even millions of dollars in pre-idea funding simply for showing promise. The result, Baker argues, is a culture that has created an “action first, question later” mindset rife with hype and even deception. The wisdom of the approach is not without its critics. Sam Altman, a Stanford dropout, calls the VC social circuit an “anti-signal” for real talent, while former Stanford president John Hennessy points out that the most successful startups have historically come from the work of graduate students, not undergraduates. For now, at least, the FOMO-driven competition for new talent means that Stanford remains one of the country’s premier proving grounds for future tech elites.