Neurodegeneration: Alzheimer's Disease

Alzheimer's disease is a progressive neurodegenerative disorder defined neuropathologically by two hallmark lesions that Alois Alzheimer himself described in 1906: senile (amyloid) plaques, extracellular deposits of aggregated amyloid-beta peptide, and neurofibrillary tangles, intracellular filaments of hyperphosphorylated tau protein. Clinically, it produces a characteristic profile of decline: episodic memory loss first (because the hippocampus and entorhinal cortex degenerate earliest), followed by deficits in language, visuospatial skill, executive function, and eventually basic self-care, over a typical course of 8–10 years from diagnosis to death. A crucial conceptual distinction runs through the whole field: AD is defined by pathology, but pathology begins long before symptoms. Amyloid deposition can be detected by PET imaging or cerebrospinal fluid (CSF) assay 15–20 years before the onset of measurable memory impairment. This preclinical phase, followed by mild cognitive impairment (MCI) and then dementia, defines a biological continuum now formalized in research diagnostic frameworks (the "ATN" system: Amyloid, Tau, Neurodegeneration biomarkers) rather than a strict clinical cutoff. The disease has two etiological forms that share final pathology but differ in cause. Early-onset familial AD (EOAD), fewer than 5% of cases, is caused by autosomal-dominant mutations in the amyloid precursor protein gene (APP) or in the presenilin genes (PSEN1, PSEN2), which encode the catalytic core of γ-secretase; onset is typically before age 65, sometimes in the 30s or 40s. Late-onset ("sporadic") AD, the overwhelming majority of cases, has no single causal mutation but is strongly modulated by the APOE gene, particularly the ε4 allele, along with age, vascular risk factors, and dozens of smaller-effect genes identified by genome-wide association studies (GWAS).

Step 1 — APP processing and the amyloidogenic pathway. Amyloid precursor protein (APP) is a single-pass transmembrane glycoprotein expressed throughout the brain, embedded in synaptic and neuronal membranes, with proposed roles in synapse formation and cell adhesion. APP can be cleaved by two competing pathways.

• In the non-amyloidogenic pathway, α-secretase (an ADAM-family metalloprotease) cuts within the Aβ sequence, precluding formation of intact Aβ and releasing a soluble, neuroprotective fragment (sAPPα).
• In the amyloidogenic pathway, β-secretase (BACE1) first cleaves APP at the N-terminus of the Aβ domain, releasing sAPPβ and leaving a membrane-bound C-terminal fragment (C99). γ-secretase — a multi-subunit intramembrane protease complex whose catalytic subunit is presenilin-1 or -2 — then cleaves C99 within the membrane, releasing Aβ peptides of variable length (mostly Aβ40, a smaller but more aggregation-prone and neurotoxic fraction Aβ42) plus the APP intracellular domain (AICD).

The Aβ42:Aβ40 ratio matters enormously: Aβ42 aggregates far faster and is enriched in plaques. Familial AD mutations in APP near the cleavage sites, or in PSEN1/2 (which shift γ-secretase's cutting register), consistently increase the proportion of Aβ42 relative to Aβ40 — direct molecular support for the idea that relative overproduction of the more aggregation-prone peptide, not just total amyloid, drives disease.

Step 2 — Aggregation and the oligomer hypothesis. Aβ42 monomers self-assemble into soluble oligomers, then protofibrils, then insoluble fibrils that deposit as extracellular plaques (often with a dense amyloid core surrounded by dystrophic neurites — "neuritic plaques"). Critically, evidence from the 1990s–2000s (notably work from Dennis Selkoe's group) showed that soluble oligomeric Aβ species, not mature fibrillar plaques, correlate best with synaptic dysfunction: oligomers extracted from human AD brain, injected into rodent hippocampus, impair long-term potentiation (LTP) and disrupt memory within hours — far too fast to reflect plaque formation. This reframed amyloid toxicity as a synaptic, not merely structural, disease.

Step 3 — Tau hyperphosphorylation and tangle formation. Tau is a microtubule-associated protein that normally stabilizes axonal microtubules, particularly in unmyelinated axons. In AD, tau becomes abnormally hyperphosphorylated (at more than twice the normal number of phosphate residues) by kinases including GSK-3β and CDK5, whose activity is thought to be promoted by upstream Aβ-driven signaling. Hyperphosphorylated tau detaches from microtubules, destabilizing axonal transport, and self-aggregates into paired helical filaments, which accumulate as neurofibrillary tangles inside the neuronal soma and dendrites.

Step 4 — Spatial and temporal spread ("tau propagation"). Tangles do not appear randomly; they follow a highly stereotyped anatomical sequence first described by Heiko and Eva Braak (the Braak staging system, 1991): starting in the transentorhinal cortex, spreading to the hippocampus and limbic structures, and finally to neocortical association areas. This staged spread appears to occur partly via trans-synaptic, prion-like propagation — misfolded tau seeds released from one neuron can template misfolding of normal tau in synaptically connected downstream neurons, demonstrated by injecting pathological tau into transgenic mouse brains and observing spread along anatomically connected pathways rather than by simple diffusion.

Step 5 — Synaptic failure and selective neuronal death. Oligomeric Aβ binds synaptic receptors (implicated targets include NMDA receptors and cellular prion protein, PrP^C, as a co-receptor), triggering excessive calcium influx, mitochondrial dysfunction, and internalization of AMPA and NMDA receptors — physically weakening synapses before any cell dies. Downstream, chronic microglial activation (attempting to clear Aβ) releases pro-inflammatory cytokines that contribute further synaptic and neuronal damage — the "neuroinflammatory" arm of pathology, genetically supported by AD-risk genes expressed almost exclusively in microglia (TREM2, CD33). Eventually neurons undergo apoptosis and necrosis, driving the atrophy visible on structural MRI.

Step 6 — The cholinergic hit. The basal forebrain cholinergic system — the nucleus basalis of Meynert, medial septum, and diagonal band — sends long, diffusely branching projections releasing acetylcholine (ACh) throughout the hippocampus and neocortex, modulating attention, encoding, and consolidation. These large cholinergic neurons are disproportionately vulnerable early in AD, partly because their long, poorly myelinated axons and dependence on nerve growth factor (NGF) retrograde transport make them metabolically fragile, and partly because they lie in cortical regions hit earliest by tau pathology. The result is a marked loss of cortical choline acetyltransferase (ChAT) activity — the enzyme that synthesizes ACh — that correlates with the severity of dementia. This single observation, the cholinergic hypothesis (Bartus et al., 1982; Davies & Maloney, 1976), is the direct pharmacological basis for the first class of approved AD drugs.

Step 7 — From synapse to network to cognition. Because the hippocampus and entorhinal cortex are hit first (Braak stages I–II) and are essential for encoding new episodic memories, the earliest clinical sign of AD is difficulty forming new memories, while old, well-consolidated memories and general knowledge (semantic memory, procedural skill) are relatively spared until later. As tangle pathology spreads to association cortices (Braak stages V–VI), language, spatial navigation, and executive control progressively fail — the temporal order of Braak spread maps remarkably well onto the temporal order of symptom emergence.

Alois Alzheimer, 1906–1907: the founding case. Alzheimer examined "Auguste Deter," a 51-year-old woman with progressive memory loss, disorientation, paranoia, and language breakdown. After her death, using Bielschowsky's silver-staining method, he described two novel lesions in her cerebral cortex: miliary foci ("plaques") and dense bundles of altered fibrils inside neurons ("tangles"). He presented this in 1906 and published it in 1907, establishing that a distinct pathological entity — not simply "senile" aging — underlay her dementia. This was purely descriptive, but it defined the disease that every subsequent mechanistic study has tried to explain.

Glenner and Wong, 1984: the molecular identity of the plaque core. Using biochemical purification of amyloid deposits from AD and Down syndrome brains (Down syndrome patients, who carry an extra copy of chromosome 21 — where APP resides — almost universally develop AD pathology by their 40s), Glenner and Wong isolated and sequenced a novel ~4 kDa peptide, which they named beta-amyloid. This was the critical link: it showed that plaques had a specific, sequenceable protein composition, not just amorphous debris, and it immediately suggested that a gene-dosage effect of APP could cause the pathology — a hypothesis confirmed a few years later when APP itself was cloned and mapped to chromosome 21.

Familial mutations, 1991–1995: causal genetic proof. John Hardy and colleagues (1991) identified the first APP mutation (the "London" mutation) causing autosomal-dominant early-onset AD. Subsequent work identified dozens of APP mutations clustering near the secretase cleavage sites, and — more consequentially — mutations in PSEN1 (1995, Sherrington et al.) and PSEN2, together accounting for most familial cases. Every single one of these mutations increases production of Aβ42 or the Aβ42:Aβ40 ratio. This convergence of independent mutations in different genes, all funneling onto the same biochemical outcome, is the strongest causal evidence for the amyloid cascade hypothesis, formalized by Hardy and Higgins in 1992: that Aβ accumulation is the initiating molecular event, with tau pathology, inflammation, and neuronal death occurring downstream.

Transgenic mouse models, 1995 (PDAPP) and beyond. Games et al. (1995) created transgenic mice overexpressing mutant human APP; these mice developed age-dependent, region-specific amyloid plaques resembling human AD pathology, along with synaptic loss and memory deficits on tasks like the Morris water maze — direct experimental proof that Aβ overproduction alone, in an otherwise normal mouse genome, is sufficient to generate AD-like pathology and cognitive impairment. Crossing APP-mutant mice with mutant PSEN1 mice accelerated plaque formation further, mechanistically confirming the γ-secretase link.

Oligomer toxicity experiments, 1998–2002 (Selkoe, Klein, Ashe labs). Rather than relying on plaques, several groups isolated soluble, non-fibrillar Aβ oligomers directly from the brains and CSF of AD patients and from cell culture. When injected into the cerebral ventricles of healthy rats, these soluble oligomers — at concentrations far below what would be needed to form visible plaques — acutely impaired hippocampal long-term potentiation and disrupted learned behavior within hours. This was decisive: it dissociated Aβ toxicity from plaque burden (many cognitively normal elderly people have substantial plaque load at autopsy) and refocused the field on soluble oligomers as the proximate synaptotoxic agent.

PET amyloid imaging in living patients: Klunk & Mathis, 2004 (Pittsburgh Compound B). Before this, amyloid pathology could only be confirmed post mortem. The development of [¹¹C]PiB, a radioligand that binds fibrillar Aβ and can be visualized by PET, allowed researchers for the first time to quantify amyloid burden in living people and correlate it longitudinally with cognitive decline — the technical foundation of the entire preclinical-AD biomarker framework and of patient selection for anti-amyloid drug trials.

Anti-amyloid immunotherapy trials, 2021–2023: the therapeutic test of the hypothesis. Monoclonal antibodies designed to bind and clear Aβ (aducanumab, then lecanemab, then donanemab) were shown by amyloid-PET to substantially reduce plaque burden in treated patients. In the Clarity AD trial (lecanemab, 2023, NEJM), patients receiving the antibody showed a statistically significant 27% slower decline on the CDR-SB (Clinical Dementia Rating–Sum of Boxes) scale over 18 months compared to placebo — a modest but real clinical effect, providing the first direct human causal evidence that lowering Aβ can slow (not reverse) cognitive decline, while also revealing side effects (ARIA: amyloid-related imaging abnormalities, i.e., brain edema/microhemorrhage) that complicate the risk-benefit calculus.

Example 1 — Interpreting a family pedigree for early-onset AD. A 38-year-old man presents with progressive memory loss. His father died of dementia at 46; his paternal grandmother had dementia in her 40s; his paternal aunt is affected at 44. His mother's side is entirely unaffected across three generations. Genetic testing shows he carries a heterozygous PSEN1 mutation; his unaffected siblings do not carry it. Reasoning: The pattern — every generation affected, transmission through both sexes, only paternal-line inheritance, full penetrance by the mid-40s — is classic autosomal dominant inheritance, consistent with a single copy of a PSEN1 mutation being sufficient to cause disease. Because PSEN1 encodes the catalytic subunit of γ-secretase, this mutation almost certainly shifts cleavage of APP toward the more aggregation-prone Aβ42, accelerating amyloid deposition by decades relative to sporadic AD. This is why familial cases present 30–40 years earlier than typical late-onset AD (~65+). Expected recurrence risk for each of his own children: 50%, independent of sex, since the gene is autosomal. This case is a direct clinical parallel to the pedigrees Hardy's and Sherrington's groups used to originally map APP and PSEN1.

Example 2 — Estimating cholinergic deficit and predicting drug response. Postmortem biochemical assay of a patient's temporal cortex shows choline acetyltransferase (ChAT) activity at 35% of age-matched control values, while cortical thickness on prior MRI was reduced by only about 8% relative to controls at the time of diagnosis. Reasoning: The disproportion between a large ChAT loss (65% deficit) and comparatively mild structural atrophy (8%) tells us that cholinergic synaptic/neurotransmitter dysfunction substantially precedes gross neuronal death in cortical terminal fields — consistent with basal forebrain cholinergic neurons losing function and axonal ACh output before the cell bodies themselves are lost, and consistent with cortical atrophy lagging behind synaptic/biochemical decline. Clinically, this predicts that a cholinesterase inhibitor (e.g., donepezil), which blocks acetylcholinesterase and raises synaptic ACh by slowing its breakdown, should produce a measurable — though symptomatic, not disease-modifying — improvement in attention and memory performance in this patient, because there is still meaningful residual cholinergic machinery (some ChAT activity, some intact terminals) for the drug to potentiate. As the disease progresses and cholinergic neurons die outright, this symptomatic benefit diminishes, which is exactly the trajectory seen clinically: cholinesterase inhibitors help most in mild-to-moderate stages and lose efficacy in severe AD.

Alois Alzheimer was a German psychiatrist and neuropathologist working at the Frankfurt asylum and later Munich, in an era when psychiatry was just beginning to seek biological substrates for mental illness rather than treating it as purely a disorder of character or environment. He was a meticulous microscopist, trained in the new silver-staining techniques developed by Max Bielschowsky, and worked in the laboratory of Emil Kraepelin, the towering figure of descriptive psychiatric classification. Alzheimer's genius was not inventing a new stain or theory, but insisting on correlating a patient's detailed clinical course — which he tracked personally for years — with precise microscopic findings after death, at a time when "senile dementia" was dismissed as an unremarkable feature of old age not worth dissecting scientifically.

Diagnosis and biomarkers. Amyloid-PET and CSF/plasma Aβ42:Aβ40 and phosphorylated-tau (p-tau181, p-tau217) assays now allow AD pathology to be detected in living patients years before dementia, driving the shift from a purely clinical diagnosis to a biological (ATN) framework — directly relevant to Brain Imaging Methods: Structural and Functional.

Symptomatic pharmacotherapy. Acetylcholinesterase inhibitors (donepezil, rivastigmine, galantamine) and the NMDA-receptor antagonist memantine remain first-line symptomatic treatments, translating the cholinergic and glutamatergic mechanisms above directly into prescribing practice — a direct extension of Drugs Acting on Neurotransmitter Systems.

Disease-modifying therapy. Anti-amyloid monoclonal antibodies (lecanemab, donanemab) are now FDA-approved for early symptomatic AD, representing the first drugs shown to alter the underlying pathological trajectory rather than only masking symptoms, alongside intensive research into anti-tau immunotherapies and BACE inhibitors (though several BACE inhibitor trials failed or worsened cognition, a sobering lesson about pathway timing and off-target effects).

Genetic counseling and risk stratification. APOE genotyping and, in familial cases, direct APP/PSEN1/PSEN2 sequencing inform reproductive counseling and enrollment in prevention trials (e.g., the DIAN and API studies following at-risk family members from young adulthood).

Regenerative and cell-based strategies. As neuronal loss becomes irreversible in later stages, stem-cell-derived neuronal replacement and trophic-factor delivery strategies are being explored to rebuild degenerated circuits — the natural bridge to Neural Stem Cells and CNS Repair.

Public health and prevention. Because vascular risk factors (hypertension, diabetes, physical inactivity) measurably influence AD risk and age of onset, dementia prevention is now a mainstream public-health target, with the Lancet Commission estimating that addressing 14 modifiable risk factors across the lifespan could prevent or delay roughly 45% of dementia cases globally.

Ask the tutor: How exactly does the Aβ42:Aβ40 ratio, rather than total amyloid, determine aggregation risk and disease onset?