Basic Science · Alzheimer’s Disease

Why Some Scientists Question the Amyloid Hypothesis — and What It Means for APOE ε4 Carriers

The debate over amyloid’s role in Alzheimer’s disease is legitimate, scientifically productive, and frequently misunderstood. This article presents both sides with equal rigor.

Brian Paquette, DO, MPH  ·  Neurology & Pain Medicine  ·  APOE ε4 Carrier  ·  Peer-reviewed sources cited throughout
The Short Version

For decades, the leading theory of Alzheimer’s disease held that a protein called amyloid-beta was the primary cause. Many drugs were developed to remove it. Most of those drugs failed to help patients, even when they successfully cleared amyloid from the brain. This has led serious researchers to ask whether amyloid is truly the cause — or simply one part of a more complex picture.

Importantly, none of this debate changes the strength of the APOE ε4 association with Alzheimer’s risk. That association holds regardless of which theory turns out to be correct — because APOE ε4 affects multiple disease pathways simultaneously.

The Amyloid Cascade Hypothesis

In 1992, researchers Hardy and Higgins proposed what became the dominant framework for understanding Alzheimer’s disease: the amyloid cascade hypothesis. The core idea is that Alzheimer’s begins when a protein fragment called amyloid-beta (Aβ) accumulates in the brain, forming plaques that trigger a downstream chain of events — including tau protein tangles, inflammation, and ultimately the death of nerve cells responsible for memory and cognition.

This hypothesis attracted enormous scientific investment and substantial supporting evidence. Rare, inherited forms of Alzheimer’s are caused by mutations in genes that directly regulate amyloid production (APP, PSEN1, PSEN2). People with Down syndrome, who carry an extra copy of the amyloid gene, develop amyloid plaques early and have high rates of Alzheimer’s disease. And the APOE ε4 allele — the strongest common genetic risk factor for Alzheimer’s — is known to impair the brain’s ability to clear amyloid.

For these reasons, the amyloid cascade hypothesis became the organizing framework for Alzheimer’s drug development for over three decades.

The Empirical Challenge: Decades of Negative Drug Trials

The strongest evidence cited by hypothesis skeptics is not theoretical — it is clinical. A succession of drugs designed to reduce amyloid in the brain failed to slow cognitive decline in large, well-designed Phase III trials, even when they successfully achieved their biological target.

Solanezumab · Anti-amyloid antibody

Failed in three Phase III trials (EXPEDITION, EXPEDITION2, EXPEDITION3) enrolling thousands of patients with mild-to-moderate Alzheimer’s disease. Despite clear evidence that the drug was binding to amyloid and reducing free Aβ in the cerebrospinal fluid, cognitive and functional outcomes were unchanged compared to placebo. A subsequent prevention trial in individuals without symptoms also showed no benefit and numerical worsening on some measures. (Doody et al., NEJM 2014; Honig et al., NEJM 2018; Sperling et al., NEJM 2023)

Bapineuzumab · Anti-amyloid antibody

Failed in two Phase III trials stratified by APOE ε4 status, enrolling over 2,400 participants. The drug reduced amyloid on PET imaging and lowered phosphorylated tau in cerebrospinal fluid — objective evidence of biological activity — yet neither cognitive nor functional outcomes improved in either carriers or non-carriers. The trials also identified a significant safety signal: brain swelling (ARIA-E) occurred at rates that limited dosing, particularly in APOE ε4 carriers. (Salloway et al., NEJM 2014)

Verubecestat · BACE-1 inhibitor

A different approach — instead of clearing amyloid, this drug blocked one of the enzymes responsible for producing it, reducing cerebrospinal fluid amyloid by 63–81%. Despite this substantial reduction, the mild-to-moderate Alzheimer’s trial showed no cognitive benefit. More concerning, the prodromal (pre-dementia) trial was terminated early: patients on the active drug showed significantly worse cognitive outcomes than those on placebo, and were more likely to progress to dementia (HR 1.38). (Egan et al., NEJM 2018; Egan et al., NEJM 2019)

Semagacestat · γ-secretase inhibitor

This Phase III trial was stopped early by safety monitors. Patients receiving the drug showed significantly worse cognitive decline than those on placebo and experienced more skin cancers, infections, and other serious adverse events. Subsequent laboratory research suggested the drug may have actually increased toxic amyloid species inside cells, rather than reducing them — a reminder that biological systems respond in complex and sometimes counterintuitive ways. (Doody et al., NEJM 2013; Tagami et al., Cell Reports 2017)

A 2026 Cochrane meta-analysis — the most rigorous type of systematic evidence review — examined all amyloid-targeting antibody trials and concluded that cognitive effects were “trivial” and functional effects “small at best,” despite successful amyloid removal from the brain. The authors stated plainly that “successful removal of amyloid from the brain does not seem to be associated with clinically meaningful effects in people with mild cognitive impairment or mild dementia due to Alzheimer’s disease.” (Nonino et al., Cochrane Database 2026)

A Note on Scientific Integrity: The Lesné Retraction

In 2022, investigative reporting in Science raised concerns about a highly influential 2006 paper by researcher Sylvain Lesné, published in Nature, which had introduced a specific form of amyloid — a protein cluster called Aβ*56 — as a key driver of memory impairment. Independent scientific analysis, subsequently confirmed by the University of Minnesota, found evidence that images in the paper had been manipulated. The paper was retracted by Nature in 2024.

This retraction received significant public attention and was interpreted by some as invalidating the amyloid hypothesis entirely. That interpretation is not scientifically accurate. The existence of amyloid plaques in Alzheimer’s brains, and their association with disease, is established by thousands of independent studies. What the Lesné paper had contributed — evidence for a specific oligomeric form of amyloid as a particular memory toxin — was one line of mechanistic research among many, not the foundation of the hypothesis itself.

What the Lesné controversy does legitimately illustrate: that peer review, while essential, is not infallible; that high-citation papers can contain errors that go unchallenged for years; and that the scientific community’s willingness to investigate and correct the record is itself evidence that the self-correcting nature of science, though slow, is functioning. It is a lesson in scientific epistemology, not a conspiracy.

What Else Might Be Driving Alzheimer’s Disease?

The failures of amyloid-only treatments have encouraged research into other biological mechanisms. These are not fringe ideas — they are actively investigated by mainstream academic neuroscience and supported by substantial evidence.

Tau protein — the more proximate driver

Tau is a protein that normally stabilizes the structural skeleton of nerve cells. In Alzheimer’s disease, tau becomes chemically altered and forms tangled masses inside neurons, eventually killing them. A landmark study of 1,431 patients directly compared how well amyloid PET scans and tau PET scans predicted cognitive decline over time. Tau PET was consistently the stronger predictor — in patients with MCI, tau PET explained 25% of the variance in cognitive decline, compared to 7% for amyloid PET. When both were measured together in statistical models, only tau contributed meaningfully; amyloid added no independent predictive value once tau was accounted for. (Ossenkoppele et al., JAMA Neurology 2021; Aschenbrenner et al., Neurology 2018)

Neuroinflammation — the immune system’s role

The brain has its own immune cells, called microglia. Normally they function as surveillance and cleanup crews, clearing debris and responding to damage. In Alzheimer’s disease, microglia become chronically activated and may drive neurodegeneration rather than simply responding to it. Critically, several of the genes most strongly associated with Alzheimer’s risk — including TREM2, CLU, and CR1 — are not amyloid genes at all. They are immune genes. This convergent genetic evidence suggests that immune dysfunction is not a downstream consequence of amyloid, but a primary contributor to disease. (Wang et al., Frontiers in Immunology 2023; Botella Lucena & Heneka, Acta Neuropathologica 2024)

Other contributing factors

Vascular disease, mitochondrial dysfunction, disrupted sleep and glymphatic clearance, metabolic factors including insulin resistance, and oxidative stress all appear to contribute to Alzheimer’s pathophysiology either independently or in interaction with amyloid and tau. Emerging research has also identified novel signaling peptides — such as T14, studied in the context of selectively vulnerable subcortical neurons — as possible upstream initiators of neurodegeneration, with amyloid and tau accumulating downstream. (García Ratés et al., Alzheimer’s & Dementia 2024; Liu et al., Redox Biology 2023)

What the Recent Positive Trials Tell Us

The scientific picture shifted meaningfully in 2022–2023 with the publication of the CLARITY AD trial (lecanemab) and the TRAILBLAZER-ALZ 2 trial (donanemab). Both drugs cleared amyloid from the brain and produced statistically significant — though modest — slowing of cognitive decline. This is the strongest clinical evidence to date that amyloid is causally relevant to Alzheimer’s disease, not merely a bystander.

Trial Drug Slowing of Decline Key Caveat
CLARITY AD Lecanemab 27% slowing (CDR-SB) Greater benefit in non-carriers and heterozygotes; homozygotes excluded due to ARIA risk
TRAILBLAZER-ALZ 2 Donanemab 35% slowing in low/medium tau group (iADRS); 22% in combined population Benefit was substantially attenuated in patients with high baseline tau burden — no significant effect on primary endpoint in high-tau group

Two features of the donanemab data deserve particular attention. First, the magnitude of benefit was strongly dependent on how much tau pathology was already present: participants with lower tau burden derived roughly 35% slowing, while those with substantial tau accumulation showed minimal benefit. Second, patients treated at the earliest stages of predicted disease progression experienced roughly 60% slowing, compared to 17% in those at more advanced stages. (Sims et al., JAMA 2023; Raket et al., Alzheimer’s & Dementia 2026)

These findings are consistent with a multi-hit model: amyloid appears to be a necessary contributor to disease, but not the sole driver. Once substantial tau pathology has developed, clearing amyloid alone is insufficient to meaningfully reverse the course. The current evidence best supports a view in which amyloid, tau, neuroinflammation, and other factors each contribute, and in which early intervention — before the cascade is fully established — yields the greatest benefit.

Why This Debate Does Not Diminish APOE ε4 Risk

Some APOE ε4 carriers reading about this debate understandably wonder: if scientists are not sure amyloid is the cause, does that mean the APOE ε4 association with Alzheimer’s is also in doubt?

The answer is no — and the reason is important. APOE ε4 does not raise Alzheimer’s risk through a single amyloid-dependent pathway. It operates across multiple mechanisms simultaneously:

  • It impairs the brain’s ability to clear amyloid-beta from perivascular pathways (relevant to the amyloid model)
  • It promotes tau hyperphosphorylation and tangle formation (relevant to the tau model)
  • It impairs microglial function and drives chronic neuroinflammation (relevant to the immune model)
  • It disrupts cholesterol transport, synaptic lipid homeostasis, glucose metabolism, and mitochondrial function — pathways that are relevant under any mechanistic framework (Liu et al., Nature Reviews Neurology 2013)

APOE ε4 is not a hypothesis-dependent risk factor. It is a pan-mechanistic vulnerability gene. Whichever theory of Alzheimer’s disease turns out to be most correct, APOE ε4 carriers are implicated through that pathway. The debate over amyloid is a debate about mechanism — not about whether carriers face elevated risk, or whether early intervention and risk modification matter.

Where the Scientific Field Stands Today

The scientific community has largely moved away from a strict amyloid-only view toward a multi-pathway model of Alzheimer’s disease. Amyloid is understood to play a real and important role — the positive lecanemab and donanemab trials confirm this — but it is increasingly seen as one component of a more complex cascade rather than the singular cause.

The key open questions are now about timing (how early must intervention begin?), about patient selection (who benefits most, and how do we identify them?), and about combination approaches (can targeting amyloid, tau, and neuroinflammation simultaneously produce better outcomes?). A large pooled study of 6,514 participants across 21 cohorts found that individuals who were positive for both amyloid and tau PET faced a 57% absolute risk of progression to MCI or dementia over five years — compared to 17% for those positive for amyloid alone. This underscores that the combination of pathologies, not either one in isolation, drives clinical risk. (Moscoso et al., JAMA 2025)

Scientific debate of this kind is not a sign of a field in crisis. It is a sign of a field advancing — re-examining assumptions, demanding better evidence, and building toward more effective treatments. For APOE ε4 carriers, the practical conclusion remains the same regardless of which mechanistic model prevails: early risk assessment, cardiovascular and metabolic optimization, sleep quality, and monitoring for emerging biomarkers all carry evidence-based rationale that is independent of any single hypothesis.

Primary Sources Referenced in This Article
  • Doody RS et al. Phase 3 trials of solanezumab. N Engl J Med. 2014
  • Salloway S et al. Two Phase 3 trials of bapineuzumab. N Engl J Med. 2014
  • Egan MF et al. Verubecestat for mild-to-moderate AD. N Engl J Med. 2018
  • Egan MF et al. Verubecestat for prodromal AD. N Engl J Med. 2019
  • Doody RS et al. Semagacestat Phase 3 trial. N Engl J Med. 2013
  • Nonino F et al. Amyloid-targeting antibodies Cochrane review. Cochrane Database. 2026
  • Ossenkoppele R et al. Tau PET vs amyloid PET prognostic comparison. JAMA Neurology. 2021
  • Aschenbrenner AJ et al. Tau PET, amyloid PET, and hippocampal volume. Neurology. 2018
  • Moscoso A et al. Tau PET positivity and clinical outcomes. JAMA. 2025
  • Sims JR et al. TRAILBLAZER-ALZ 2 (donanemab). JAMA. 2023
  • Raket LL et al. Donanemab treatment by baseline tau burden. Alzheimer’s & Dementia. 2026
  • Perry R et al. Lecanemab in ApoE ε4 heterozygotes/non-carriers. J Prev Alzheimers Dis. 2026
  • Liu CC et al. ApoE and Alzheimer disease: risk, mechanisms and therapy. Nat Rev Neurol. 2013
  • Wang C et al. Microglia-associated neuroinflammation in AD. Front Immunol. 2023
  • Botella Lucena P, Heneka MT. Inflammatory aspects of AD. Acta Neuropathol. 2024
  • García Ratés S et al. Novel neuronal mechanism upstream of amyloid. Alzheimers Dement. 2024

This article is written for educational purposes and does not constitute medical advice. Content is authored by a licensed physician and grounded in peer-reviewed literature, but individual clinical decisions should be made in consultation with a qualified healthcare provider familiar with your specific circumstances.