There's a man in Wales who earns millions of dollars by reading other people's research. He sits down with papers that have already passed peer review, already been printed in the biggest journals in the world, and he goes looking for the mistakes everyone else missed [1].
He's good at it. Last December, a major cancer institute in Boston paid out fifteen million dollars to settle a case built on data problems. His share, as the whistleblower, was 2.6 million dollars [2].
His name is Sholto David. And a few weeks ago he pulled on a loose thread that concerns the longevity and anti-aging community — the exact product code of an antibody, the little tool labs use to detect a specific protein. It's a detail buried in the methods section that most readers skip.
By the time he finished pulling, he'd found something that should raise a serious red flag if you've ever bought a supplement that promised to slow down aging. Hundreds of studies across this entire field may have been measuring the wrong molecule, for more than a decade, without anyone noticing. And the supplements built partly on top of that research were already outrunning their evidence.
Table of Contents
- What the Detective Found
- The Zombie Cells, and the Magic Bullet
- It Worked in Mice — and Why That Part Survives
- The Test Nobody Markets
- The Trials That Were Never Proof
- The Bottom Line
- References
What the Detective Found
So here's what David actually did. He sat down with a stack of aging-research papers — 334 of them that he could fully read — and he checked the one detail almost nobody reads: the catalog number of the antibody each lab had bought to find their target protein [3].
The protein everyone was hunting is called p16. Think of it as a warning flag. When a cell gets old and stops working properly, it tends to wave this flag — so for years, if you wanted to know whether a treatment was clearing out old, broken cells, you measured p16. More flags, more old cells. Fewer flags, the treatment worked [4].
But here's the problem David found. There's a second, completely unrelated protein that happens to share almost the same name — p16-ARC. Different gene. Different job. It helps build the cell's internal scaffolding. It has nothing to do with aging [3].
And here's what makes the mix-up so easy: type "p16" into a major antibody catalog, and the first product that comes up isn't the aging protein at all — it's p16-ARC. So lab after lab, reaching for an off-the-shelf p16 antibody, clicked the top result and walked away with a reagent for the wrong protein entirely [5].
In 312 of those 334 papers, the researchers had bought antibodies for the wrong one. They thought they were measuring the aging flag; they were measuring the scaffolding protein instead [5]. And this error didn't hide in obscure journals — it showed up in Nature, Nature Medicine, Cancer Cell and others, the most prestigious publications in science, for over a decade [5].
David's own read is that in many cases it looks more like an honest mistake than fraud. But a few of these papers reported results that only make sense if you'd used the right antibody, which raises an uncomfortable question about how the wrong one ended up in the methods [5]. One cancer biologist put it bluntly: this is "an indictment … of the ability of people to conjure stories based on flawed data" [5].
The Zombie Cells, and the Magic Bullet
Here's why p16 matters so much. In 1961, a scientist named Leonard Hayflick was growing human cells in dishes, and he noticed something nobody expected. The cells divided — a few dozen times — and then they just stopped. They didn't die. They didn't keep going. They sat there, alive but frozen, refusing to divide again [6].
This was heresy. The entire establishment believed normal cells could divide forever, given the right conditions. That belief rested on the most famous cell experiment of its time — Alexis Carrel's chick-heart cells at the Rockefeller Institute, kept alive and dividing, supposedly, for over thirty years. It was seen as proof that aging wasn't driven by something happening within cells [7][8].
So Hayflick was picking a fight with a legend. And he lost, for a while — the journal that first received the paper rejected it outright [9]. But Hayflick was right. Cells do have a limit. And Carrel's "immortal" culture, kept alive for thirty-four years, has never been fully explained.
Decades later, a researcher named Judith Campisi showed those retired cells aren't just sitting there quietly. They pump out a stream of inflammatory signals that irritate and damage the healthy tissue around them [10].
Informally, these senescent cells became known as zombie cells: cells that won't divide, haven't died, and quietly poison their neighbours. And the suspicion grew that these zombie cells weren't just a feature of aging — they might be one of its causes [11].
Now you can see why p16 matters. Zombie cells look almost identical to ordinary cells from the outside, so the field needed a flag to spot them — and the flag it settled on was p16. The very same protein now sitting at the centre of Sholto David's scandal. And if p16 could mark the enemy, the obvious next move was to kill it.
It Worked in Mice — and Why That Part Survives
In a lab at the Mayo Clinic in Rochester, Minnesota, a researcher named Jan van Deursen and his team did something clever — something that, crucially, didn't rely on any antibody at all. Instead of trying to spot the zombie cells from the outside, they rebuilt the mice from the inside. They added a piece of genetic code that only switches on inside a cell that's already waving the p16 flag, and attached to that code was a self-destruct command.
So the cell itself becomes the detector. There's no guessing, no staining, no antibody that might be reading the wrong protein. Give the mouse one harmless trigger drug, and every cell with the code switched on quietly clears itself out, while every healthy cell is left untouched.
And the results were remarkable. The mice that had their zombie cells cleared out stayed healthier; their tissues held up better [12]. In a follow-up, clearing those cells actually extended how long the mice lived [13].
Now this part is crucial, because it helps keep in perspective what the scandal touches and what it doesn't. Van Deursen's kill switch was genetic — it targeted the cells through the gene itself and a drug trigger. There was no antibody involved. So the strongest evidence that clearing senescent cells actually helps an animal does not depend, even a little, on the broken antibody David found.
That's why careful aging scientists — Matt Kaeberlein among them — have been saying the antibody mistake doesn't debunk the underlying biology. And on that, they're right. The proof that zombie cells matter comes from these genetic experiments and other lines of evidence, not from the molecule that got mislabelled [14].
So if the genetic experiments survive, what exactly did David's scandal break? This. For years, the case for zombie cells wasn't just those few clever mouse experiments. It was a mountain of papers — hundreds of them — reporting that these cells pile up in aging tissue, in one disease after another. That mountain is what made the whole field feel like settled science, solid enough to build an industry on top of. And a huge part of that mountain was measured with the wrong tool.
The careful genetic work — the foundation — still stands. But a lot of what got stacked on top of it was never measuring the right protein at all. And the longevity and anti-aging supplements that rode all of that excitement? They were sold on the scenery.
Here's what happened. In 2015, a group led by James Kirkland coined a word for drugs that kill zombie cells: senolytics [15]. And out of all the compounds that could work this way, one stood out — not because it was the most powerful, but because anyone could buy it. Fisetin: a cheap plant compound, sold in a capsule, available without a prescription. Early mouse studies looked spectacular, with better function and longer life [16].
That single 2018 paper, more than any other, is what launched a thousand supplement labels. A pill that clears your zombie cells — it sold itself. But spectacular-in-mice is where a lot of ideas go to die, so eventually someone ran fisetin through the one test built to be impossible to fool.
The Test Nobody Markets
It's called the Interventions Testing Program, run by the National Institute on Aging, and it is brutal by design. It runs the same compound in three separate labs at the same time — Jackson Laboratory in Maine, the University of Michigan, and a centre in San Antonio, Texas — with around 150 mice in each group, all following the exact same strict protocol. The mice themselves are deliberately bred to be genetically varied, so a result can't be a fluke of one inbred strain. No single lab can quietly nudge the outcome in a preferred direction [17].
That 2018 fisetin paper had been tested in groups roughly thirty times smaller than this — which is exactly how an effect can look spectacular in a small study and then evaporate in a big one [18]. So the Interventions Testing Program ran fisetin. And here's what happened: nothing. Fisetin did not extend lifespan — not in the males, not in the females, not on either dosing schedule they tried [18].
And then comes the part that ties this whole story together. When the program checked whether fisetin had actually cleared any zombie cells, they didn't use the broken antibody. They measured p16 a different way — by counting the gene's activity directly, with a method called qPCR, the one approach that sidesteps David's entire scandal. Again, the result was negative: fisetin made no measurable difference. The senescent cells were real — p16 climbed with age, just as the theory says it should — but fisetin simply didn't clear them [18].
The Trials That Were Never Proof
It's no wonder, then, that the human trials have been disappointing. Take the famous 2019 study of dasatinib and quercetin — a senolytic combo — in people with diabetic kidney disease. It included just nine people and lasted three days, with no placebo group [19]. That's not a trial that tells you whether a drug works; it's a trial that tells you whether you can run a bigger one.
The pattern repeats. An Alzheimer's trial of the same combo reported that cognitive and neuroimaging endpoints did not significantly differ from baseline [20]. The lung-fibrosis studies people point to follow the exact same shape — the first was fourteen patients, open-label, with no comparison group, designed only to check whether the drugs were tolerable [21]. When researchers finally ran a small randomised version a few years later, the conclusion wasn't that the drugs worked — it was, once again, that the approach seemed feasible and a bigger trial was needed [22].
And then there's the one human study that wasn't a pilot at all: a randomised, double-blind, placebo-controlled trial — the real thing — testing fisetin in seventy-four people with knee arthritis, using the same intermittent senolytic dosing the mouse studies used. This was fisetin's best shot at proving itself in people, under proper conditions. The result: no meaningful improvement in pain, in function, or in the joint itself, compared to a placebo [23].
The Bottom Line
Notice the pattern. The pilots only ever checked that it was safe enough to keep going — and the one time fisetin got a proper, randomized, placebo-controlled test, it came up empty. After years of headlines, there is still not a single trial showing these supplements deliver a real health benefit in humans.
The idea might still pay off one day. The biology of clearing zombie cells is real and worth chasing, and somewhere down the line a properly run human trial might surprise us. But this is what it looks like when a field's marketing gets ahead of both its measurement and its evidence at the same time. The pattern is always the same — striking results in mice, a cheap pill you can buy today, and a human trial that's coming soon. Now you know the shape of it, so you can spot it the next time a capsule promises to do what only a real study can prove.
References
1. https://www.statnews.com/2024/01/27/sholto-david-profile-dana-farber-retractions/
3. https://forbetterscience.com/2026/06/02/mind-over-antibody/
4. https://www.jci.org/articles/view/22475
6. https://www.sciencedirect.com/science/article/abs/pii/0014482761901926
7. https://embryo.asu.edu/pages/hayflick-limit
8. https://www.researchgate.net/publication/250781436_Unlike_ageing_longevity_is_sexually_transmitted
9. https://garfield.library.upenn.edu/essays/v7p218y1984.pdf
10. https://pmc.ncbi.nlm.nih.gov/articles/PMC2592359/
11. https://discoverysedge.mayo.edu/2023/11/16/health-and-zombie-cells-in-aging/
12. https://www.nature.com/articles/nature10600
13. https://www.nature.com/articles/nature16932
14. https://x.com/mkaeberlein/status/2066193950306713709
15. https://pmc.ncbi.nlm.nih.gov/articles/PMC4531078/
16. https://pmc.ncbi.nlm.nih.gov/articles/PMC6197652/
17. https://pmc.ncbi.nlm.nih.gov/articles/PMC6241422/
18. https://pmc.ncbi.nlm.nih.gov/articles/PMC10828146/
19. https://pmc.ncbi.nlm.nih.gov/articles/PMC6796530/
20. https://pmc.ncbi.nlm.nih.gov/articles/PMC10168460/
21. https://pmc.ncbi.nlm.nih.gov/articles/PMC6412088/
22. https://pmc.ncbi.nlm.nih.gov/articles/PMC10006434/
23. https://www.oarsijournal.com/article/S1063-4584(25)00692-2/abstract
