Alzheimer's disease

Alzheimer's disease is the neurodegenerative disease behind 60–70% of dementia: a progressive loss of memory, then language, judgement, and finally the capacity to live unassisted, marked in the brain by extracellular plaques of amyloid-β and tangles of tau protein inside neurons. Its overwhelming risk factor is age. For the arguments this wiki tracks it is the counterpart of Cancer: cancer sets the ceiling on how long a body lasts, and Alzheimer's asks whether the mind lasts as long as the body does.

§An age-related disease

Dementia prevalence roughly doubles with every five years of age past 65, making it an old-age disease in the demographic sense cancer is. The World Health Organization estimates that 57 million people were living with dementia in 2021, with nearly 10 million new cases a year, and counts it the seventh leading cause of death worldwide; the Alzheimer's Association estimates 7.2 million Americans aged 65 and over have Alzheimer's dementia in 2025, with a projected near-doubling by 2060 absent better treatment or prevention.¹ The growth is driven by population aging; age-specific risk has actually declined in several high-income countries.

Why this matters here

Eliminating dementia would add little life expectancy, because its victims are old and die of something else soon after; what it would remove is a large share of late-life disability. That makes Alzheimer's the sharpest test of the distinction between lifespan and Healthspan, and the geroscience reading, that it is a manifestation of brain aging rather than a disease that happens to strike the old, is one of its central claims. Any intervention that extends the body without protecting the brain extends the years in which this disease can arrive.

§Plaques, tangles, and the amyloid hypothesis

Alois Alzheimer described the disease from a single autopsy: Auguste Deter, who died in 1906 in her mid-fifties after five years of progressive memory loss and delusions, and whose cortex showed the plaques and neurofibrillary tangles that still define the diagnosis.² The plaques are aggregates of amyloid-β, a peptide cleaved from the amyloid precursor protein; the tangles are hyperphosphorylated tau. Both are misfolded-protein assemblies that spread between cells in a templated, prion-like manner, which is why the disease is read as the clinical face of Proteostasis collapse failure. Around that core sit most of the other Hallmarks of aging: chronically activated microglia tie the disease to Inflammaging, senescent glia accelerate pathology in mice, and disturbed sleep precedes diagnosis by years. Whether the adult hippocampus makes new neurons at all is the dispute covered under Stem cell exhaustion.

The field's organizing idea, stated by Hardy and Higgins in 1992, is the amyloid cascade hypothesis: amyloid-β accumulation is the initiating event, and tau pathology, inflammation, synapse loss, and dementia follow downstream.³ The strongest evidence has always been genetic. The hypothesis's persistent embarrassment is anatomical: tangle burden tracks symptoms far better than plaque burden does, and a substantial minority of cognitively normal older people carry a plaque load that would support a diagnosis.

1. 1906
   Auguste Deter's autopsy
   Alois Alzheimer presents the case of a patient in her fifties with progressive dementia, describing the plaques and tangles found at autopsy; the written report follows in 1907.
2. 1984
   Amyloid-β identified
   George Glenner and Caine Wong purify and sequence the amyloid-β peptide from cerebrovascular deposits, giving the plaque a molecular identity.
3. 1991
   First causal mutation
   Alison Goate and colleagues report a mutation in the amyloid precursor protein gene that causes early-onset familial Alzheimer's.
4. 1992
   The amyloid cascade hypothesis
   John Hardy and Gerald Higgins propose in Science that amyloid deposition initiates the disease and everything else follows.
5. 1993
   APOE ε4 linked to common disease
   Corder and colleagues show that risk of late-onset Alzheimer's rises with each inherited copy of the APOE ε4 allele.
6. 2021
   Aducanumab approved, then abandoned
   The FDA grants accelerated approval over its advisory committee's objection; Biogen discontinues the drug in early 2024 after minimal uptake.
7. 2023
   Lecanemab's traditional approval
   CLARITY AD reports a 27% slowing of decline over 18 months, the first clearly positive phase 3 result for an anti-amyloid antibody.
8. 2024
   A hard year for amyloid's reputation
   Nature retracts the 2006 Aβ*56 paper for doctored images; donanemab is approved in the US; the Lancet Commission names 14 modifiable risk factors.
9. 2025
   Diagnosis from a blood draw
   The FDA clears the first blood test for Alzheimer's pathology, a plasma p-tau217 to amyloid-β ratio, for symptomatic adults 55 and over.

§Genetics

Mutations in three genes (APP, PSEN1, and PSEN2, all altering amyloid-precursor processing) cause aggressive early-onset disease with near certainty, but account for on the order of 1% of cases or fewer. A rare APP variant that reduces amyloid-β production is associated with protection, which is among the stronger genetic arguments for the cascade. It is also the variant He Jiankui later announced he wanted to install in embryos, a proposal condemned like any Human germline editing plan built on one association.

The common-disease genetics run through APOE. One ε4 copy roughly triples risk; two copies raise it by around an order of magnitude, and a 2024 multi-cohort analysis argued that ε4 homozygosity should be considered a distinct genetic form of the disease outright: nearly all homozygotes showed amyloid pathology in biomarker studies by age 65, and while they are about 2% of the population they account for an estimated 15% of cases.⁴ The claim is contested, since many homozygotes die without dementia, but it has practical force: ε4 carriers suffer the highest rates of the new antibodies' signature side effect. Early-phase gene therapy trials aimed at the APOE locus are underway.

§The amyloid controversy

Between the cascade hypothesis and the first approval lie almost three decades in which every amyloid-directed drug failed, a record critics read as the hypothesis tested and found wanting, and defenders as trials run too late, on the wrong patients, with drugs that never engaged the target. Rival framings (tau-first, vascular, inflammatory, infectious) attracted attention in rough proportion to amyloid's setbacks. A 2022 investigation then found doctored images in a highly cited 2006 paper on the amyloid oligomer Aβ*56; Nature retracted it in June 2024.⁵ The fraud touched one strand of oligomer work, not the genetics or the trials, but it sharpened the complaint that amyloid absorbed funding the alternatives never saw.

Is amyloid the cause?

Proponents point to the causal genetics and to the fact that the only drugs ever to slow the disease are the ones that clear plaques. Critics reply that the slowing is modest, that plaque removal is nearly complete while the clinical effect is not, and that tau tracks symptoms far better — consistent with amyloid being an upstream trigger, a partial cause, or a tombstone. The prevention trials now dosing amyloid-positive people before symptoms are the closest thing to a decisive experiment either side has proposed.

§The antibody era

Aducanumab's 2021 accelerated approval, granted on plaque clearance over its advisory committee's objection, ended as a cautionary tale: insurers balked, and Biogen discontinued the drug in January 2024. What replaced it is a genuine but modest result, twice replicated. In CLARITY AD, lecanemab slowed decline on the CDR-SB scale by 27% over 18 months (a difference of 0.45 points on an 18-point scale) and won traditional US approval in July 2023.⁶ In TRAILBLAZER-ALZ 2, donanemab slowed decline by about 35% in the low-and-medium-tau population and was approved in July 2024, with dosing stopped once plaques clear.⁷ Both drugs cause amyloid-related imaging abnormalities, or ARIA: brain swelling in 12.6% of lecanemab recipients and 24% of donanemab recipients — usually asymptomatic, though three deaths in the donanemab trial were attributed to it. Both require proven amyloid pathology, repeated MRI monitoring, and carry US list prices near $30,000 a year before the cost of scans and infusions.

Slowing is not stopping

Treated patients in both trials continued to decline: the drugs bought back part of 18 months' worsening, not a plateau. The between-group differences sit below the one-point CDR-SB threshold many clinicians had used for a minimally important difference, and whether the benefit compounds, holds, or fades over five or ten years is unknown; the extension data so far come from sponsors, against external cohorts rather than placebo. Mainstream opinion spans "first disease-modifying success, build on it" to "marginal benefit at real risk and cost", and both positions cite the same trials.

Repurposing hopes have fared worse: two large trials of oral semaglutide in early Alzheimer's, the most prominent test of the idea that GLP-1 receptor agonists might protect the aging brain, reported no slowing of decline in late 2025.⁸

§Diagnosis moves to blood

For most of its history Alzheimer's could be confirmed only at autopsy, then by PET scans and lumbar punctures too costly and invasive to scale. In May 2025 the FDA cleared the first blood test for the underlying pathology, a plasma ratio of p-tau217 to amyloid-β 1-42 that tracks amyloid PET closely, for use in symptomatic adults aged 55 and over, explicitly not as a screening test.⁹ Cheap blood biomarkers are quietly redefining the disease as a biological state that begins a decade or more before symptoms, the same preclinical framing the Aging biomarkers field wants for aging itself, and they raise the question that hangs over Consumer blood testing: what a positive result is worth when intervention is unproven. Prevention trials dosing amyloid-positive, cognitively normal volunteers with the approved antibodies are underway, with results expected late this decade.

§Prevention without a prescription

The 2024 Lancet Commission attributes around 45% of dementia to 14 modifiable risk factors across the life course — from low education, hearing loss and hypertension to the two added in 2024, high midlife LDL cholesterol and untreated vision loss.¹⁰ The estimate is a theoretical ceiling built from observational associations, assuming each factor is causal and could be fully eliminated, and no trial has prevented Alzheimer's with any of it. It is still the only large claim on offer, and it points the same way as the falling age-specific incidence in some rich countries: dementia risk moves with education, blood pressure, exercise and cardiovascular health before any drug enters the picture.

§What it means for longevity arguments

Alzheimer's is where Compression of morbidity stops being an abstraction: added years spent with dementia are precisely the outcome that argument exists to rule out. Several interventions this wiki tracks have been tested against it, so far without clinical success: a Senolytics pilot showed only tolerability, young-plasma trials feasibility with debated efficacy, and a Memory prosthesis presumes circuitry the disease removes. The antibodies bound the arithmetic from the other side: the best therapy slows one disease of brain aging by about a third, at real risk, nowhere near the repeated, compounding rescue Longevity escape velocity requires. For projects that treat the brain as the thing worth preserving, from Brain preservation outward, Alzheimer's is the demonstration that the substrate can fail first, slowly, while the body lasts. The open question is whether removing amyloid before symptoms prevents the disease rather than slightly delaying it; the prevention trials now running are the first able to answer.

§See also

• Cancer
• Proteostasis collapse
• Geroscience hypothesis
• Inflammaging
• Compression of morbidity
• Aging biomarkers
• Parabiosis and young blood
• Memory prosthesis

§References