Avi Roy

A weekly jab in the belly is generating more revenue than the entire AI industry. Ozempic + Mounjaro: $71B in 2025. OpenAI + Anthropic: $29B. And they've barely started. ~2% of the 800 million eligible patients can currently access them. h/t @DrSamuelBHume

@DevinNeko In this case, the compensation ran through a second inflammasome, NLRP1, rather than TGF-β. TGF-β is a major aging signal on the fibrosis and immune-aging side, just not the pathway that backfired in this study.

@agingroy Maybe TGF-β signaling?

Every cancer has to do two things. Keep growing, and never die. A protein called c-Myc is how most of them do the growing. It’s running too hot in roughly 70% of human cancers, telling the cell to divide, and divide, and keep dividing long after a healthy cell would have stopped. Push it high enough, though, and the cell kills itself. Biologists have stared at that contradiction since 1992. A preprint from the National Cancer Institute has an answer. c-Myc clumps. The protein normally floats around loose, and when a cell makes too much of it, it starts sticking to itself into a solid clump, the same process that builds the plaques in an Alzheimer’s brain, and the clump is what sets off the cell’s suicide. The authors think that’s the point, a growth signal that carries its own off switch. They found the same clumps in Alzheimer’s brains. Nobody knows yet what they’re doing there.

The training, fasting, and walking are what actually drive change, so lead with those. Two caveats on the rest. GABA isn't a growth hormone precursor; it's a neurotransmitter. It may nudge GH release a little in a lab, but the effect is small, and nobody has shown it does anything you'd notice. DHEA keeps coming up empty in elderly trials too, with nothing on strength, metabolism, or mood once you account for the hormones it turns into. Save your money for sleep and the barbell.

@agingroy If you want to approach it careful how about going low dose and paired with intermittend fasting (24h/wk), weight lifting, walking, DHEA (for men depending on status), GABA (hGH precursor) and 100% bioavailable full spectrum amino acids?

Ozempic and Wegovy are often talked about as vanity drugs. In reality, they represent one of the biggest shifts modern medicine has made in a generation, and almost nobody describes them that way. For decades, every blockbuster appetite drug was eventually pulled for harming people. Fen-phen damaged heart valves and was banned in 1997. Sibutramine increased heart attacks and strokes and was gone by 2010. Lorcaserin was pulled in 2020 after a cancer signal. The pattern seemed fixed: if a drug was strong enough to switch off hunger, it was strong enough to hurt you. Then semaglutide, the molecule in Wegovy, did the opposite. In a trial of 17,604 people who were overweight or obese and had heart disease but not diabetes, it cut heart attacks, strokes, and cardiac deaths by 20%. A sister trial slowed kidney disease by 24%. For the first time, a weight-loss drug protects organs instead of damaging them. That's why the knock-on effects are showing up everywhere, from women's employment to a £780m drop in UK supermarket spending, which this @FT Economics Show episode explores. Obesity is no longer a willpower story. It's a treatable disease.

You're onto something with the network point, and it's That’s exactly why this NLRP3 result hurts. They knocked down one node, and the system rerouted through a second inflammasome, NLRP1, creating more inflammation, not less. Single targets are brittle because the network pushes back. Where I'd slow you down is on the exosome part. Signaling is the right layer to target, but exosome therapy is still mostly promise; the human data are thin, and the field runs hot on hype. The diagnosis is good. The tool isn't ready yet.

I think these are remnants of “stopping /slowing down aging is more doable and sensible target than reversal of aging” mindset. What ages us is ,acc to my limited readings, is gradual deterioration of signaling efficiency in repair networks which in turn gives way to aging process in an accumulated manner. working on signaling (exosome therapies) at this stage leading to age reversal is both more doable an more impressive; counterintuitively, i think.

For a decade, the anti-aging bet was to lower NLRP3, a protein that switches on inflammation and helps drive age-related disease. A new paper says lowering it partway made mice age faster. The logic was sound. Chronic inflammation rises with age and contributes to heart disease, Alzheimer's, diabetes, and frailty, and NLRP3 is a major source. Removing it entirely in mice makes them age more slowly. That result, from 2013, kicked off a race to drug NLRP3. But a drug doesn't delete NLRP3. It only lowers it. So @CorderoMarioD's team made mice with one working copy instead of two, which is closer to what a drug does. Those mice aged faster: shorter lives, more frailty, worse metabolism, and, in females, earlier ovarian failure. The mice compensated. Lowering NLRP3 caused a second inflammasome, NLRP1, to rise and form a hybrid complex that produced more inflammation, not less. Cutting both restored them. Those blockers are already in clinical trials in people. Roche, NodThera, Ventyx, and Novartis all have NLRP3 inhibitors in human trials for conditions including Parkinson's, obesity, and heart disease, and each one lowers the protein rather than deleting it, the exact move that backfired in these mice. If that carries over, a partial blocker could worsen the inflammation it targets unless it also hits NLRP1. Still, it's mice, it's genetics not a drug, and human NLRP1 biology differs. A flag, not a verdict. Thursday, 16th July, @LauraMinquini and I will cover it with the author @CorderoMarioD on @athenabiorg's Menopause and Fertility show. The same inflammation that ages the body also ages the ovary.

Then we agree, because that's not what I'm saying. A range isn't a wall. Your genes deal the opening hand; they don't set a date you can't pass. And the data leans your way here. Once people get past 105, their yearly odds of dying stop climbing and flatten out (Barbi, Science). No hard ceiling has shown up yet. In animals, we already push the maximum with drugs like rapamycin. So treat that 55% as a target. It's the part we haven't learned to move yet, and moving it is the whole job. That's the opposite of fatalism.

@agingroy @NewYorker @DhruvKhullar 1. Thanks for all the time you have invested in this discussion. Much appreciated. 2. You are raising fair points. 3. I refuse to accept that my body is programmed to live a predetermined maximum number of years and that it's impossible to go beyond that. That's too fatalistic.

We keep hunting for the village that cracked living to 100. The bigger news is that the village may not exist. In @NewYorker, physician @DhruvKhullar reviews Saul Newman’s book Morbid, which makes a deflating case. The world’s “blue zones” line up almost perfectly with the places that have poor birth records, deep poverty, and a history of pension fraud. Fix the paperwork and the miracle fades. When US states brought in birth certificates, their counts of people over 110 dropped by up to 82%. Okinawa, the famous one, has Japan’s worst obesity rate, and it’s the KFC capital of the country. What lengthens a life is dull and mostly out of your hands. The richest Americans outlive the poorest by more than a decade, and that gap keeps widening. Your lifespan tracks where you live more than what you eat.

Up to a third of the weight you lose on Ozempic isn't fat. It's muscle. That's a problem, so drug companies raced to fix it, and they're winning. Add a muscle-saving drug to a shot like Ozempic and the weight you lose flips to about 92% fat, with less than half the muscle loss. @EliLillyandCo's bimagrumab already does this, and @ScholarRock and @Regeneron are close behind. Wall Street analysts at TD Cowen think that market alone hits $30 billion by 2035. And that's only the start. Fat loss was the first frontier, and it's basically won, with even stronger drugs coming. Muscle is the second. The third is sleep. New drugs can flip the brain's wake switch, turning alertness up in the morning or sleep on at night, close to on demand. Lilly just paid $6.3 billion for @centessa to lead it. Fat, muscle, sleep. @EliLillyandCo is worth a trillion dollars, the most valuable drug company alive, and it's betting on all three. We spent a decade learning to shed fat. Next, we learn to keep our muscles and switch sleep on and off.

This Thursday, 16 July, on @BIOHBN. 2 PM ET, 8 PM CET. It's the Menopause and Fertility show, hosted by @LauraMinquini, who built @athenabiorg to fund women's health science that almost no one else will. I'm co-hosting. Three guests worth the hour: @CorderoMarioD, the scientist behind the paper above, on the drugs that could extend fertility and the ones that might backfire. @kutlukoktay, who pioneered ovarian tissue freezing, on preserving fertility itself, and why HRT alone isn't enough. @rivatez, whose essay "Against Maximum Stimulation" argues for gentler fertility medicine over max-dose egg freezing and IVF. The ovary is the first organ to age. This is the room working out what to do about it. RSVP on @BIOHBN luma.com/egeqxhbm

Sources & Notes: The paper: Muela-Zarzuela, Cordero et al., Science Advances, 26 June 2026. doi.org/10.1126/sciadv… The history. In 2000 Claudio Franceschi named "inflammaging," the slow, sterile inflammation that rises with age and feeds heart disease, dementia, diabetes and frailty. The NLRP3 inflammasome, a danger sensor inside immune cells, became the prime target. In 2013 Vishwa Dixit's lab (Youm et al., Cell Metabolism) deleted NLRP3 in mice and they aged better. doi.org/10.1016/j.cmet… That result launched a drug race. The new twist. A knockout removes NLRP3 entirely; a drug only turns it down. This paper tested the drug-like state, one gene copy, about half the protein. Those mice aged faster: median survival about 23 months vs 30 for controls, hair loss 62% vs 30%, spinal curvature 68% vs 9%, corneal clouding 51% vs 6%, plus worse metabolism and, in the females, lower AMH, higher FSH and fewer ovarian follicles. Knock down NLRP1 as well and lifespan returned to about 28 months. The drugs it flags, all partial NLRP3 blockers now in human trials: MCC950 (halted in phase 1 for liver injury), selnoflast (Roche, early Parkinson's), NT-0796 (NodThera, obesity), VTX2735 (Ventyx, recurrent pericarditis), DFV890 (Novartis). The caveats: it's mice; it's a gene deletion, not a drug, so "inhibitors backfire" is an inference the authors did not test directly; and humans carry one NLRP1 gene where mice carry several, so the compensation may not map cleanly. What does travel: the team showed NLRP1 and NLRP3 physically bind in human cells (THP-1), so the hybrid is not a mouse-only artifact.

Fair, the question everyone really cares about is this: how much of your lifespan is actually up to you? You may have heard the figure of 80% lifestyle and 20% genes. It appears behind every supplement and morning routine, but the problem is where it came from. That number treated every death the same way. Cancer at 75, a car crash at 30, and a bad infection were all lumped together. And you're right that anything can happen to anyone. A bus doesn't check your DNA. But that's the catch. Those random deaths have almost nothing to do with your genes, and keeping them in the same pile is what made genetics look small. Uri Alon's team at Weizmann, writing in Science, removed accidents and infections and asked a cleaner question. Of the people dying because the body simply wore out, how much of the timing is in the genes? The answer roughly doubled, from about 20% to about 55%. They also checked it against twins raised apart, so it holds up. That's where you and I actually meet. Your genes give you a range, maybe you're built to reach 78, maybe 98. What you do: the sleep, the food, not smoking, decides where inside that range you land. Lifestyle is huge. So is the hand you were dealt. Neither wins alone. And no contradiction with my first post. Doubting a village that claims a pile of 110-year-olds with no birth certificates is about sloppy record-keeping. It was never me saying you can't change how you age.

@agingroy @NewYorker @DhruvKhullar With all due respect, don't you think that what you've just written contradicts your 1st Tweet? I also beg to disagree with the claim that "what lengthens a life is dull & mostly out of your hands." Anything can happen to anyone at any moment, but lifestyle beats genetics.

Half of the people over 60 in the West have diverticular disease, which causes small pouches to bulge through the colon wall. By 85, it’s closer to two-thirds. When one becomes infected or bursts, you get fiber, antibiotics, and if it perforates, emergency surgery. There is no drug that changes the disease. For fifty years, the explanation has been that you didn’t eat enough fiber. A genetic study, still a preprint, compared 114,806 people who have it with 819,224 who don’t, and found 236 regions of the genome, 126 of them new, pointing to 214 genes. Those genes aren’t about digestion. They sit in the smooth muscle that squeezes the colon, the connective tissue holding its wall together, and the interstitial cells of Cajal, which are the pacemaker cells that set the rhythm of the squeeze. The colon has a pacemaker, and this disease appears to be written into it.

The first issue is the trap itself: without a patent, no one ever recoups the cost of a trial, so it only gets done if someone runs it without a profit motive, through philanthropy, the NIH, or a nonprofit that simply publishes the results. No one has built that for peptides yet. With the kits, it’s the same gap from the other side. A lower price per vial doesn't matter much when no one has tested what's in it, so you're dosing on trust, with no purity or safety data behind it.

@agingroy The issue is who pays for the trials when there’s no patents on it?

Everyone wants to sell peptides, but nobody will fund even a basic clinical trial on them. Wonder why?

Agreed the field is real, but I'd drop the centenarian count as evidence for it. Saul Newman's work, the one that won an Ig Nobel last year, found that only 18% of "validated" supercentenarians even have a birth certificate, and the records cluster in poor regions with no registration, which points to pension fraud and clerical error more than biology. The case for longevity medicine doesn't need those records anyway. It rests on aging itself being modifiable, which rapamycin, GLP-1s, and reprogramming are already showing. Still WHEN not IF, just not because of the record books.

@agingroy @NewYorker @DhruvKhullar 2/2 Just because a family defrauded the government to exploit the system doesn't mean people cannot live beyond 100. (Dhruv Khullar's The New Yorker article) People have lived beyond 100, and they will live even longer. It's a matter of WHEN, not IF.

Agreed, and what still gets overlooked is that obesity sits upstream of most of that list, so we can finally treat the common cause instead of managing ten complications one at a time. SELECT already showed the ripple effect, with a 20% reduction in cardiac events just from treating obesity. The "normalize it" framing was never going to hold up against a drug that measurably lowers risk.

@agingroy Many people tried to normalize obesity, but obesity increases risk for hypertension, diabetes, heart disease, stroke, cancer, arthritis, infertility, bad pregnancy outcomes, and bad infectious disease outcomes. Finally having effective medicines to treat obesity is a huge win.

@ItsJakePerry Ha, perfect. Though it's almost the reverse of Pied Piper here, everyone selling peptides already has the revenue, they just can't afford the one trial that might prove their bestseller doesn't work. Russ would tell them to stay pre-evidence.

@agingroy Reminds me of this m.youtube.com/watch?v=BzAdXy…

Milk does raise IGF-1 a little, so the instinct isn't crazy. But that doesn't translate into more cancer when you look at the outcomes. The largest pooled review (WCRF, 111 cohorts) found that higher dairy intake was linked to less colorectal cancer, about 13% lower per 400g a day. Prostate cancer is the one that leans the other way, and only mildly. I wouldn't cut out a whole food group to avoid a growth signal that small.

@agingroy Should ppl stopconsuming dairies therefore? Bc it activates growth in body?

It roughly does, with one caveat. The latest measured NHANES, 2021 to 2023, came in at 40.3%, down from 41.9% the cycle before, the first time in about a decade the measured line hasn't gone up. It's not a statistically significant drop on its own, so I'd take it with a grain of salt. But it lines up with your point: if the self-report gap is steady, then the slope is the signal, and this is the first time the measured and self-reported curves are both tipping down together.

@agingroy My guess is the self-report gap stays pretty flat year to year, so a falling line still means something even if the level's off. Curious if the first measured NHANES read agrees.

Bariatric being the exception makes sense, since it's the one line where the drop is already visible in this quarter's numbers. The rest moves more slowly, which is why nobody's modeling it. A heart attack that doesn't happen never shows up as lost revenue; it just isn't there, so cardiology won't feel SELECT-sized effects until years after they begin. By the time it becomes legible in a budget, the volume is already gone.

@agingroy @OmVenture The indirect cost are likely more than that. Yet I can tell you that in conversation after conversation with boards and C-execs, I have never encountered *any* that are thinking about any of this seriously — other than bariatric surgery, where volumes are already falling.

I think the stats point the other way here. A confidence interval that's narrow and sits entirely below 1 is the stronger result; it means the effect is measured precisely, not inflated. HR 0.80, 0.72 to 0.90. And the "is it the drug or the weight" question mostly goes away in a randomized trial, because the only thing that differs between the two arms is who got semaglutide, so the weight loss is one of the drug's effects. The event curves also split before most of the weight came off, which argues against it being just the weight.

@agingroy I am a proponent of GLP-1s, but the MACE data is WAY FUCKING OVERSTATED. The CI doesn't even cross one and is incredibly tight. There's no way to say it's even from the drug itself and is likely more related to the weight loss itself.

Exactly, that's the reason. The M is c-Myc, the same oncogene the post is about, so it's the last thing you want switched on in a cell you're trying to make younger instead of cancerous. Sinclair's group dropped it and used just OSK in the retina, and they still got the methylation age to run backwards and brought vision back in old and glaucomatous mice (Lu, Nature 2020). You give up a little reprogramming speed and take the tumor risk way down.

@agingroy And a reason to use OSK and not OSKM?

Cheaper than the excuse suggests. A bare-bones 100-person trial runs about $1M. BPC-157 sells for $150 to $400 a month, so that whole trial is what a few hundred of your own customers spend on it in a single year. The RECOVERY trial enrolled 47,000 people and rewrote COVID care worldwide on a £2.1M grant, about $2.7M. The cash is already in the till. Nobody wants to spend it finding out the answer might be no.

@agingroy Cost to run studies very expensive