OZEMPIC / BIOMIMETICS
What Ozempic's side effects say about the future of medicine
Ozempic — and, in general, the class of drugs known as GLP-1 agonists — might be the most significant medical breakthrough since anti-retroviral therapy. Just a few years ago Novo Nordisk was a run-of-the-mill Danish pharmaceutical company producing anti-diabetic medicine. Then, in 2021, a study showed that a higher dose of that medicine also caused weight loss. In less than three years after that, Novo Nordisk became the most valuable company in Europe, bypassing Shell and LVMH. Eli Lilly, a US company producing similar drugs, is also doing great.
Why lifelong obesity drugs would become such a market darling requires no explanation. But I think there’s something else that’s interesting about them.
Let’s look at the list of top 10 bestselling drugs in 2023 and how they work at the molecular level.
Keytruda (pembrolizumab). It is a drug that targets various cancers. It works by blocking a receptor on an immune cell that allows other body cells to tell the immune cell “don’t kill me”. Cancers use this receptor to prevent the immune cells from attacking it. Blocking the receptor makes the immune cells more ruthless, and they go on to attack the cancer.
Humira (adalimumab). It is a drug against various autoimmune diseases such as arthritis, psoriasis and Crohn’s disease. Inflammation can sometimes ramp up to a point that it becomes a self-sustaining process which cannot resolve, causing damage to the body. Part of how inflammation is sustained is via a molecule called TNF-α, which is released into the site of the inflammation and acts as a signal for immune cells to continue wreaking havoc. The drug blocks TNF-α and breaks the loop of self-sustaining inflammation.
Ozempic (semaglutide). Ozempic works by imitating the gut-derived hormone glucagon-like peptide 1 (GLP-1), which is released when we eat and triggers the feeling of satiety. It basically makes the body think “I’m already full”.
Eliquis (apixaban). It is an anticoagulant that prevents blood clots and reduces the risk of stroke. Blood coagulation is a chain reaction which involves multiple enzymes acting upon each other, eventually causing platelets to clump and create a clot. The drug blocks one of the enzymes in the chain reaction and stops the process.
Dupixent (dupilumab) — like Humira, this is also a drug against autoimmune diseases. This one blocks not TNF-α but another inflammation-sustaining molecule, interleukin-4. More specifically, it blocks the receptor to that molecule, rather than the molecule itself (so, rather than stopping a radio transmission that says “continue inflammation”, the drug breaks the antennas on cells that receive this transmission).
Biktarvy (bictegravir/emtricitabine/tenofovir alafenamide). This is a cocktail of three drugs against HIV. Two of them block the enzyme that the virus uses to replicate its genome, and another one blocks the enzyme that copies and pastes that viral genome into our own — this integration is part of what makes retroviruses so dangerous.
Comirnaty (tozinameran). This is an mRNA vaccine against COVID-19. Vaccines work by showing the immune system some element of the virus so it can develop protection against the virus. This would naturally happen upon infection, but with a vaccine you can make it happen before actually encountering the virus.
Stelara (ustekinumab). Another immune suppressant, which also acts by blocking interleukins, and this time not the receptors, but the signaling molecules themselves, like Humira with TNF-α.
Darzalex (daratumumab). Chemotherapy against a blood cancer known as multiple myeloma. It is a cancer of white blood cell progenitors, and they have a cell surface molecule called CD38, which they need in order to actually become blood cells. The drug blocks this molecule.
Eylea (aflibercept). It is a drug against age-related macular degeneration. Part of the reason it happens is overgrowth of blood vessels in the eye. The abnormal growth of these blood vessels is sustained by yet another signaling molecule called VEGF-A. The drug blocks this molecule.
Notice something interesting? Almost all drugs in this list break something. They bind to some molecule that is doing its job, and block it. Only two of them instead promote something that the body does naturally: a vaccine, and Ozempic.
What makes Ozempic so successful is that it emulates the human body. Vaccines do, too: a vaccine itself is, at its core, a very simple technology that takes advantage of almost magical abilities of our immune systems to recognize and fight any pathogen. There are other examples, too: birth control pills, for instance, are basically Ozempic for sex hormones, making the body think “I’m already pregnant”. In the future, there are sure to be more medicines that will use the body’s own devices to improve health — gene therapy being the prime candidate. It seems self-evident that future medicine will reprogram biology, rather than interfere with it. With GLP-1 agonists, we may be at the precipice of this transition.
A few weeks ago I wrote about the rise of NVIDIA, the tech giant that owes its dominance to a bet it once made on GPUs — microchips that happened to emulate human brains better than traditional CPUs. Here, I think we are seeing something similar: in the end, emulating nature just works better than anything else.
One interesting aspect of both these transitions — from CPUs to GPUs and from “interference drugs” to “reprogramming drugs” is the nature of side effects. Ten years ago, computers had a totally different set of flaws than they do now: they were precise, unerring, but they only understood algorithmic commands. When we imagined artificial intelligence, we thought it would remain very machine-like — we did not anticipate that it would instead have basically the same flaws as us: it’s unreliable, prone to bias, tends to imagine things that are not there, jumps to conclusions too fast. In a way, these problems are much more natural than the problems of a traditional digital computer.
The side effects of Ozempic are also more “natural”: they themselves resemble a cohesive biological condition, rather than a wrench thrown into an intricate mechanism — the typical side effects of, say, chemotherapy. Consider a recent paper that shows how GLP-1 agonists engage, on top of satiety circuits in the brain, a separate circuit for nausea — if you are an animal that just swallowed a larger meal than you can digest, it makes sense that your body would want you to throw up. Or consider the fact that Ozempic also has mental effects, reducing sex drive and reward-seeking behavior (including drug use) and making people feel tired — it makes sense that your priority after slaying a large beast, and successfully consuming it, is to rest. This reminds me of our paper from 2017 about “food coma” in the sea hare Aplysia: we showed that in response to a meal, sea slugs produce a hormone that makes them even more sluggish, but at the same time helps them form long-term memories. This last part also explains the positive effects of GLP-1 agonists on neuroplasticity and neuroprotection: part of the “I’m full” program is to allocate energy towards forming stable memories about how you got so full in the first place.
With AI, we thought we were going from bleep-bloop straight to Skynet, but so far AI feels something like an omniscient college student. With medicine, we thought we would soon be able to reprogram our bodies, GATTACA-style, but so far the best reprograming we can figure out is to induce a food coma.
As time goes on, AI will probably seem less and less like a human and more and more like something completely new altogether. But it will probably still run on brain-like microchips. And I think the same will happen with medicine: we will eventually learn to make very specific changes of the sort that nature never envisioned. But I bet that we will still use our own physiology to achieve that.
Bottom line? Invest in biomimetics.