Will you live to 120? · Folding and Falling Apart

The most celebrated AI in all of biology

Now the chapter’s hard, real-stakes uncertainty — and it is, by design, the book’s most important one, because more readers will face this disease in someone they love than any other in these pages.

For roughly thirty years, one idea dominated Alzheimer’s research with a force few hypotheses ever achieve: the amyloid cascade hypothesis, which held that the accumulation of amyloid-beta is the primary cause of Alzheimer’s, the first domino, with tau tangles, inflammation, and neuron death all following downstream. It was a powerful and well-motivated idea — and the strongest evidence for it is genuinely compelling, so let me state it fairly before complicating it. The genetics point hard at amyloid: every known mutation that causes the rare inherited, early-onset form of Alzheimer’s affects the production of amyloid-beta; people with Down syndrome, who carry an extra copy of the gene that makes amyloid-beta’s parent protein, develop Alzheimer’s almost universally and early; and the main genetic risk factor for ordinary Alzheimer’s affects how amyloid is cleared. Amyloid is not an innocent bystander. It is causally entangled with the disease.

So the field made the obvious prediction — remove the amyloid and you should stop the disease — and spent decades and tens of billions of dollars building drugs, mostly antibodies, to clear it from the brain. And for most of those decades, the result was a graveyard that should sound familiar after the last chapter. Drug after drug succeeded in clearing amyloid from patients’ brains — they hit the target, they moved the biomarker — and failed to improve patients’ memory or slow their decline. One after another, anti-amyloid drugs cleared the plaques and left the patients no better. It was the predictor-versus-lever problem from Chapters 1 and 2, and the biomarker trap from those chapters, playing out at the scale of an entire field and a generation of patients: amyloid was a thing you could measure and a thing you could move, and for thirty years, moving it did not move the disease. The low point came when a regulator approved an anti-amyloid drug largely on the grounds that it cleared amyloid — the biomarker — over the objections of its own advisors who doubted any real benefit to patients; the drug was mired in controversy and ultimately commercially abandoned. The episode was the surrogate-endpoint cautionary tale of Chapter 4 made horribly concrete: a drug approved for moving a marker, with the thing that actually mattered — whether patients got better — left unproven.

The story grew darker still in 2022, when an investigation raised serious concerns that images in a highly influential 2006 paper — one that had pointed to a specific amyloid species as a cause of memory loss and had shaped years of research — had been manipulated. (The paper was later retracted.) It is important to be precise about what this does and doesn’t mean, because it is easy to over-read. The alleged manipulation damaged one influential thread of the amyloid story and rightly shook confidence; it did not, by itself, overturn the amyloid hypothesis, which rests on a much broader base of genetic and biological evidence than any single paper. But it is a sobering reminder, in the spirit of this whole book, that even a field this large and this well-funded can be distorted by a single compelling-but-flawed result, and that prominence is not the same as truth.

And then — the twist that makes this the book’s most genuinely nuanced case rather than a simple Chapter 5–style debunking. In the last few years, the newest anti-amyloid antibodies finally did what thirty years of predecessors could not: in large trials, they slowed cognitive decline by a statistically significant amount. Two such drugs have now been approved. This is real, and it partially vindicates the hypothesis — removing amyloid does help, which means amyloid really is causally involved, not merely correlated. But read the result with the discipline this book has tried to build, because the same trials carry the other half of the truth. The benefit is modest — a slowing of decline, not a halt and certainly not a reversal, and experts genuinely debate whether the size of the effect is large enough for patients and families to notice. The drugs carry real risks, including brain swelling and small bleeds visible on MRI that are occasionally serious. They are expensive, burdensome, and work mainly in early disease. So the honest verdict holds both halves at once: after thirty years and a graveyard of failures, clearing amyloid was shown to help — a little — which means the hypothesis was neither wrong nor right in the simple way, but partially correct and badly oversold. Amyloid is part of the story; the modest size of the benefit from removing it is the clearest possible evidence that it is not the whole story — that tau, inflammation, vascular damage, and metabolic failure are all in the mix, and that Alzheimer’s is a genuinely multifactorial, still-contested disease.

This is the place in the book where the discipline matters most, because the stakes are highest and the marketing — of drugs, of supplements, of “brain health” everything — is loudest. Anyone who tells you Alzheimer’s is solved, or that it is definitely just amyloid, or that the new drugs prove amyloid was a dead end, or that some supplement prevents it, is overclaiming in one direction or another. The truth, as of now, is harder and more honest: amyloid matters, removing it helps modestly, the disease is more complex than any single protein, and the field is still genuinely uncertain. A reader who learns only a confident headline — in either direction — will misread the news, and possibly the medical choices, for the rest of their life.

The most celebrated AI in all of biology

All 10 chapters of Will you live to 120? · Foundations · All courses