Astral Codex Ten

Alzheimer’s Disease

15 min read

Alzheimer’s Disease

Article

Alzheimer’s Disease is a recurring concept in the Astral Codex Ten archive, appearing 3 times across 3 issues between February 29, 2024 and August 14, 2025. The archive places it in contexts such as “cadaver-to-human transmission of Alzheimer’s Disease”; “In the case of Alzheimer’s research”; ""a human who had Alzheimer’s disease (Lane 3)"". It most often appears alongside aducanumab, Alzheimer’s, APP.

Metadata

  • Category: Concepts
  • Mention count: 3
  • Issue count: 3
  • First seen: February 29, 2024
  • Last seen: August 14, 2025

Appears In

Source Context

Recovered passages from the original issue text. When the raw archive preserved outbound links inside the source passage, they are listed directly under the quote.

Links For February 2024
February 29, 2024 · Original source
24: The first case reports of cadaver-to-human transmission of Alzheimer’s Disease. In the mid-20th-century, the standard treatment for dwarfism was ground up pituitary gland from the brain of a dead person. Scientists have now found that dwarfs who got pituitaries from dead people with Alzheimers developed very early Alzheimers themselves, often in their 40s. I already knew Alzheimers involved misfolded proteins, so this shouldn’t have been surprising, but I still somehow failed to think of it as a prion disease. This shows that misfolded proteins are sufficient to cause Alzheimers (with a 30 year delay? Sure, I guess, maybe that’s how long it takes a prion to spread). I’m not sure what’s left of the Alzheimers origin debate. Should we just assume that this is a protein that tends to go prion-y after enough time, or is there more to discover?
Inline links: The first case reports of cadaver-to-human transmission of Alzheimer’s Disease
Your Review: Of Mice, Mechanisms, and Dementia
July 11, 2025 · Original source
What the Heck is Going on in Panel d? Panel d is an immunoblot (Western blot), a technique that tells us whether a specific protein is being produced in a sample. The figure compares human amyloid precursor protein expression in samples from three brains, corresponding to the three “lanes” shown along the top: a normal mouse (Lane 1), a mouse with the human APP transgene (Lane 2), and a human who had Alzheimer’s disease (Lane 3). The blots (the bands and blobs) are the data. For purposes of this figure, proteins can differ in two important ways: They can differ in molecular weight (size), scored on the scale along the vertical axis on the left.
They can be expressed to a greater or lesser degree, reflected in the intensity (darkness and spread) of the bands. Lighter means less protein expression, darker more. (Note: It’s fine that there are two bands in lane 3). Let’s pause to consider what we should expect to see in this blot if the PDAPP mouse is to be considered a reasonable model of human Alzheimer’s disease. Two things! Ideally, lane 1 (normal mouse) will be empty; and
The PDAPP mouse sample in lane 2 will have the same molecular weight (height on vertical axis) and expression level (intensity and spread) as the human sample in lane 3. Why do we want to see these results? It matters because our ultimate goal is to develop treatments that work in people, not just in mice. To make that leap from a mouse model, the PDAPP mouse needs to replicate the key features of Alzheimer’s disease in humans—not just produce APP but drive similar amyloid-induced disease processes in the brain. Ok, so on point one, ✔ — nothing in lane 1. Point two? ❌ While the human sample has distinct bands in lane 3, the PDAPP mouse in lane 2 appears as a giant, smeared blob. Are these the same protein size? Impossible to tell! This happens because, relative to the human sample, the PDAPP mouse is drowning in APP—at least 10 times more, according to the paper’s text, and possibly much more by eye. When you do a Western blot, you set an exposure time for your image, just like with a manual camera: too short an exposure, and faint bands won’t appear; too long, and strong signals become an oversaturated mess. Here, no single exposure could produce similar-looking PDAPP and human samples. It’s like trying to take a photo of a candle next to the sun—you can’t adjust for both at once. A proper Western blot should show clean bands to confirm protein size and check for unexpected degradation products, but this overloaded mess makes it impossible to tell whether APP is being processed normally. A Western blot like the one shown in Panel d usually indicates either sloppy technique (overloading the gel, overexposing) or a fundamental issue with the model itself (massive expression differences between samples). The fine print explains why: the PDAPP mouse carries 40 copies of the APP transgene, all inserted at a single site in the genome. For context: ✔ At most, humans have 2 copies of APP (one from each parent). ✔ PDAPP mice have 40 human copies—plus their 2 normal mouse copies. I’m sure this blot led to high-fives in the lab—earlier models struggled to express APP at all, so getting massive overexpression must have felt like a breakthrough. But now I’m worried. If we’re trying to create a human-comparable Alzheimer’s model, this much APP might be way too much. Why might this be a problem? APP expression at this level doesn’t mirror expression levels in human Alzheimer’s. Alzheimer’s patients don’t have 40 copies of APP. If it takes this much overexpression (and a mutant form at that!) to drive pathology in mice, are these mice even an appropriate animal model for Alzheimer’s?
In Defense Of The Amyloid Hypothesis
August 14, 2025 · Original source
Overproduction or reduced clearance due to microbial infection. Amyloid-β appears to be an antimicrobial peptide and will form plaques in response to infection. [2, 3] This explains various observations that have been used to support the “infectious hypothesis”, sometimes proposed as an alternative to the amyloid hypothesis. However, it can only explain a subset of cases and, as I argue below, is even then still mediated by amyloid via an “IATN” pathway: infection → amyloid → tau → neurodegeneration. In cases of increased production, cerebrospinal fluid (CSF) will show elevated amyloid. In cases of reduced clearance, amyloid will decrease in CSF. In all cases, however, PET scans will show elevated brain amyloid, usually at first mainly in “intrinsic connectivity networks” such as the default mode network [14–20], which experience brain activity even at rest. These neurons are the most active - which causes more production and possibly less opportunity for clearance - so they tend to be the first to suffer from a production/clearance imbalance. Over time, amyloid pathology spreads spatially throughout the brain. [14, 18] Aggregations of amyloid peptides induce more such aggregations. Some of our clearest evidence for this comes from growth hormone deficiency patients, who used to have cadaver-derived ground-up brain matter injected into their own brains to provide the missing hormones. If the ground-up brain matter was sourced from the corpse of an Alzheimer’s patient, the growth hormone deficiency patients would themselves develop Alzheimer’s at a young age, probably through prion-like spread of the misfolded proteins. [21, 22] After ∼15 years of preclinical spread, the pathology eventually covers the whole brain. [14, 18] While some subtle cognitive impairment may occur during this time, it is usually not severe enough to be clinically detectable from amyloid alone. Indeed, in both humans [23–30] and mice [31–35], the severity of neurodegeneration and cognitive deficits is not a good spatiotemporal match for the severity of amyloid pathology (rather, it is a good match for the severity of tau pathology; see next section for more). These facts are often suggested as evidence against the amyloid hypothesis. However, amyloid is causally upstream of tau, as I will argue below. Therefore, the existence of cognitively normal individuals with amyloid pathology is expected in the ATN model - but typically only for a few decades, before progression to the next stage. 2: Tau pathology (T) and neurodegeneration (N) Tauopathies are a range of prion-like diseases involving the tau protein [36], whose usual function is to assist in stabilizing microtubule structure. In a tauopathy, the tau protein misfolds, and induces other, nearby tau proteins to misfold into the same shape. [37–46]. Injecting nothing but misfolded tau fibrils into a mouse brain can recruit the endogenously-produced mouse tau into this pathology, which spreads far beyond the injection site, causing neurodegeneration wherever it goes. [35, 47–59] There are at least eight distinct ways the tau protein can misfold in human disease [36], and over a dozen distinct human tauopathies, each involving a specific one of those misfoldings. These include chronic traumatic encephalopathy, Pick’s disease, corticobasal degeneration, progressive supranuclear palsy, and Alzheimer’s disease, with the last by far the most common. Each of these five diseases has its own distinct tau fold. Most normal human beings eventually develop some tau pathology in adulthood, originating probably in the locus coeruleus [60–62], which is part of the brainstem. By middle age, some amount has usually spread to the hippocampus and entorhinal cortex in the medial temporal lobe, regions responsible for episodic memory. This is called primary age-related tauopathy (PART) [63], and has its own tau fold which is distinct from most tauopathies, but the same as Alzheimer’s. [36, 64] Usually, its local severity is mild and it doesn’t spread much beyond those regions. But with sufficient amyloid pathology, this “normal” tau pathology tends to both locally worsen and spread through the rest of the brain [65], becoming the tau pathology of Alzheimer’s. Some genetic risk factors such as ApoE, in addition to affecting the clearance of amyloid-β, also increase the brain’s susceptibility to this A → T pathology conversion [66, 67]. But this is a matter of degree, as sufficient amyloid pathology seems to virtually guarantee the transition: Every 10-centiloid increase in amyloid pathology for a cognitively normal individual increases by 2.7x the probability of a PET scan detecting pathological levels of tau within five years [68]. The only known cases where patients with extremely high amyloid levels can go significant amounts of time without developing tau pathology are a few individuals with extremely rare protective genes, known only from a few case studies, e.g. [69]. Even in these instances, the individuals will eventually succumb to the tau phase, suffering neural atrophy and dementia. [70] After it forms, the tau pathology no longer appears to require amyloid’s assistance to keep spreading (although amyloid may still accelerate it). This probably explains why existing anti-amyloid therapies have been only ∼30% effective in test patients, who are usually late in the amyloid → tau progression even if early in having symptomatic disease. Neurodegeneration follows tau pathology extremely closely in time and space, in humans as well as basically all animal models, and cognitive impairments match the functions of the affected regions. There are rare reports of advanced tau pathology without cognitive decline, often in people with protective ApoE2 alleles [71], but even then, systematic analysis finds that actual density of tau inclusions is highly predictive of cognitive impairment, and that these exceptional cases usually involve widespread but locally sparse pathology [66]. The regional distribution of tau pathology explains why the first symptom of Alzheimer’s is typically impaired memory; the first cortical sites affected are usually in regions involved in memory formation. As the pathology spreads, further regions are affected, until eventually all cognitive functions are affected. As with most other aspects of the disease, the high-level picture seems relatively clear but the exact cellular and molecular pathways are not well understood (though may involve an assist from the innate immune system, especially microglia and astrocytes. [13, 35, 72]) Early Alzheimer mouse models were amyloid-only, with extremely heavy overproduction of Aβ, much more than required to recapitulate the human disease, and apparently enough to cause detectable cognitive dysfunction. However, normal mice do not get age-related tauopathy, so an amyloid-only mouse model - while useful for investigating certain questions - is not a full Alzheimer’s disease model. Combined amyloid+tau pathology mouse models, which are transgenically modified and/or injected with misfolded human tau fibrils, display the property that the presence of amyloid pathology induces the worsening and spreading of tau pathology. This is also observed in vitro in human cells. How do we know the amyloid causes the tau? Researchers have measured the correlation in many ways, from the spatiotemporal timeline (tau pathology only begins locally worsening and spreading outside the medial temporal lobe once amyloid reaches sufficient severity) [65], [98], to causal mediation modeling in the human disease [26], [99–101], to causal intervention using in vitro human cell studies [54, 102] and animal models [35, 55], [103 – 113]. But also, giving people drugs that reduce amyloid levels also decreases tau pathology. [78, 80, 82] (I’ve left out or merely alluded to much other complexity, involving the innate immune system, lipid processing, and detailed molecular and cellular mechanisms, preferring to focus on the parts of the story which are crucial to deciding the causal role of amyloid, and for which I am aware of a satisfactory account from the literature. But I don’t intend to leave the impression that the above is all there is to Alzheimer’s disease, or that all cases progress in the same exact way.) The mechanistic claims I make the following two claims about amyloid-β’s role in Alzheimer’s: Amyloid deposits are a necessary (i.e. but-for) cause in all instances of Alzheimer dementia. That is, if someone has PET or CSF positivity for amyloid and tau pathologies, and the tau pathology involves the Alzheimer tau fold and made its first cortical appearance in the medial temporal lobe, and then they developed medial temporal volume loss + amnestic mild cognitive impairment + later dementia, then counterfactually, early enough (probably ∼15 years before clinical presentation) causal intervention solely to remove the amyloid deposits would have prevented almost all tau pathology and symptoms.
Inline links: below, probably through prion-like spread of the misfolded proteins, 10-centiloid, have been, the presence of amyloid pathology induces the worsening and spreading of tau pathology
I am not claiming that environmental factors such as microbial infection can never be a cause of Alzheimer’s disease. However, I claim that in all such cases, they act upstream of the above-described disease process, inducing the formation of amyloid deposits leading to the disease. Furthermore, these cannot be all cases - there are instances of the disease in which amyloid deposits arise essentially entirely due to genetic factors (like the autosomal-dominant cases of overproduction).
Donanemab in phase 2 [81] (32%) and phase 3 [82] (35%). There have also been earlier antibodies that saw only failure in phase 3 – bapineuzumab [83, 84], crenezumab [85], solanezumab [86–88], and gantenerumab [88, 89]. These failed drugs didn’t just do a bad job treating Alzheimer’s. They also did a bad job clearing amyloid plaques, so their failure is consistent with the amyloid hypothesis. That said, just coupling the older, previously-unsuccessful antibody gantenerumab with a BBB-crossing mechanism produced extremely good target engagement and better safety in early clinical trials [74–76]. This makes me optimistic about a future BBB-crossing lecanemab (or similar), especially if given in the preclinical disease phase prior to significant tauopathy. Each of the “successes” have shown about 25-30% slowing of decline over 18 months. Some object that this isn’t clinically meaningful because it’s only a slowdown of ∼0.5 points on an 18-point CDR-SB scale, but they don’t mention that the participants start about 3 points from a perfect score (since these are relatively early-stage patients) and worsen by ∼1.5 points in those 18 months when on placebo. A literally perfect drug - one which halted all further clinical progression - could therefore only achieve about 1.5 points of efficacy on that scale. The cruxy question is whether the drugs maintain a 30% reduction after 18 months. Preliminary signs from lecanemab’s and donanemab’s open-label extensions show that they do [90], so this would amount to about 40% more years of life at each disease stage. But why have amyloid antibodies only achieved about 30% efficacy so far? The likely answer: mainly because they were given too late to prevent the downstream tau pathology cascade, but also because some of their side effects, like when they target amyloid-bearing blood vessels rather than brain tissue, can themselves worsen cognition. That said, even achieving 30% efficacy proves that amyloid plays some causal disease role and isn’t merely a downstream, harmless pathology. Why is the amyloid hypothesis unpopular? The amyloid hypothesis remains popular in the Alzheimer’s disease research community, but most press coverage is negative. These challenges are understandable, and some of them make good points, but overall fail to address the evidence discussed above. Failures and perceived failures of amyloid therapies I discussed this above, but to recap: Early attempts had suboptimal epitopes which didn’t successfully engage their targets.
Inline links: above