The Amyloid Wars Reignite: A Cochrane Review and the Unresolved Crisis at the Heart of Alzheimer's Research

There is no controversy in contemporary neuroscience that more precisely illuminates the political economy of scientific knowledge production than the Alzheimer's amyloid hypothesis debate. On 16 April 2026, STAT News reported that a new Cochrane review — the closest thing biomedicine has to a gold-standard systematic evidence synthesis — had reignited what researchers bitterly call the "amyloid wars": the decades-long dispute over whether the accumulation of amyloid beta plaques in the brain causes Alzheimer's disease, or whether it is a correlate, a consequence, or an epiphenomenon of underlying processes that remain poorly understood. The timing is significant. Lecanemab (Leqembi) and donanemab — anti-amyloid antibodies recently approved by the FDA — have entered clinical practice on the explicit claim that removing amyloid from the brain delays cognitive decline. The Cochrane review raises systematic questions about whether the evidence supporting that claim meets the evidentiary standards the field has traditionally demanded.

The amyloid hypothesis has dominated Alzheimer's research for more than thirty years. It emerged from the observation that amyloid beta peptides accumulate in the brains of Alzheimer's patients, and from genetic evidence linking mutations in the amyloid precursor protein gene to early-onset familial Alzheimer's disease. The hypothesis was elegant, it was mechanistically plausible, and it generated a research program — and a drug development pipeline — of extraordinary scope and expense. It also generated a series of clinical failures so consistent and so costly that by the early 2020s, no major therapeutic hypothesis in the history of medicine had produced so many billion-dollar drug trials that failed to benefit patients.

What Cochrane Reviews Are — and Why This One Is Contested

The Cochrane Collaboration produces systematic reviews that apply rigorous methodological standards to the totality of published evidence on clinical questions. A Cochrane review finding that anti-amyloid therapies do not convincingly demonstrate clinical benefit — or that the magnitude of benefit is too small to outweigh the risks of brain microhemorrhage that these drugs carry — would constitute a significant challenge to the scientific legitimacy of the approvals granted by the FDA.

Thomas Kuhn's account of scientific paradigms captures something important about the current moment: a paradigm in crisis does not yield to anomalies easily. Each failed amyloid trial has been interpreted not as evidence against the hypothesis but as a problem of patient selection (too late stage), dosing, formulation, or measurement. The approval of lecanemab in 2023 and donanemab in 2024 — both under the FDA's accelerated approval pathway, using amyloid clearance as a surrogate endpoint rather than clinical cognitive outcomes as the primary measure — represented a regulatory decision to bet on the hypothesis continuing to hold, against a mounting anomaly record that would, in other scientific fields, have prompted more fundamental reconsideration.

Sheila Jasanoff's analysis of regulatory science as a site where knowledge and power co-produce each other is directly applicable here. The FDA's willingness to approve anti-amyloid therapies using surrogate endpoints reflects not purely a scientific judgment but a sociotechnical imaginary: a shared vision in which Alzheimer's disease is fundamentally an amyloid problem, in which clearing amyloid must, in principle, produce clinical benefit, and in which the regulatory and scientific communities are therefore justified in acting on incomplete clinical evidence.

The Data Quality Scandal That Shadows the Hypothesis

The amyloid controversy has been further destabilized by a data integrity scandal that STAT and other science journalists have documented extensively: a seminal 2006 paper by Sylvain Lesné, which appeared to provide particularly compelling evidence for a specific amyloid oligomer's role in causing memory loss in mice, has been credibly challenged on image manipulation grounds. Lesné's affiliation with the University of Minnesota and the specific claims of that paper — which generated enormous citation influence on the research programs that followed — have been the subject of journals' expressions of concern and ongoing investigations.

This context matters for interpreting the Cochrane review's significance. The amyloid hypothesis did not rest on Lesné's paper alone; the genetic and neuropathological evidence for amyloid's relevance to at least some forms of Alzheimer's disease is real. But the specific mechanistic claims about which amyloid species cause disease, through which pathways, and at which stages of the disease course are considerably less settled than the drug development investment in amyloid-targeting therapies would suggest.

Bruno Latour's concept of inscriptions — the ways in which scientific claims are stabilized through their recording in papers, patents, approval documents, and clinical guidelines — is useful here. The amyloid hypothesis has been inscribed so thoroughly into the regulatory and clinical infrastructure of Alzheimer's medicine that challenging it now is not merely a scientific act but a challenge to an entire sociotechnical edifice, with consequences for drug approvals, insurance coverage decisions, patient expectations, and the careers of researchers whose life's work has been organized around the hypothesis.

What the Evidence Actually Shows for Approved Drugs

The clinical evidence for lecanemab and donanemab, such as it is, shows statistically significant but clinically modest slowing of cognitive decline on rating scales in early-stage patients — alongside a meaningful rate of amyloid-related imaging abnormalities (brain microhemorrhages and edema) that in some patients are asymptomatic but in others produce clinically significant effects. Translating "statistically significant slowing on a rating scale" into "clinically meaningful benefit for patients and families navigating Alzheimer's disease" requires assumptions about what scale points mean in terms of daily function, quality of life, and caregiver burden that have not been transparently established.

Donna Haraway's framework of situated knowledge asks: from whose perspective is the evidence being evaluated? Patients in early-stage disease and their families may weigh modest cognitive preservation differently from health economists calculating cost-effectiveness. Neurologists with decades of clinical experience treating Alzheimer's may evaluate brain microhemorrhage risk differently from drug approval committees under political pressure to show progress on a disease affecting tens of millions of people worldwide.

The Cochrane review's function is to synthesize this evidence systematically, without the commercial interests that shape industry-sponsored trial reporting or the institutional pressures that shape regulatory decisions. That it has reignited controversy rather than settled it says something important about the epistemic state of Alzheimer's science: the evidence base, even after enormous investment, is not yet sufficient to decisively confirm or refute the therapeutic value of the amyloid-targeting approach.

The Research Resources Distorted by Paradigm Lock-in

Steve Shapin's work on the social history of scientific institutions asks who benefits from the maintenance of particular research paradigms. The amyloid hypothesis, whatever its ultimate scientific fate, has been extraordinarily good for the pharmaceutical industry — not because the drugs have worked but because the hypothesis has sustained a drug development pipeline that generates revenue from clinical trial contracts, laboratory reagents, biomarker testing, and investor capital even when the drugs fail. Novo Nordisk's recent OpenAI partnership, cited this week, represents a bet on AI-accelerated drug discovery that will inevitably work within existing hypotheses rather than questioning them.

The researchers and clinicians working on alternative hypotheses — tau pathology, neuroinflammation, synaptic dysfunction, vascular contributions to dementia — have operated in an amyloid-dominated funding environment for decades. Whether the Cochrane review's conclusions are sufficient to shift the research resource allocation toward these alternatives, or whether the embedded institutional investments in amyloid paradigm will sustain it through another cycle of clinical development, is ultimately not a scientific question. It is a political economy question — about who controls funding decisions, whose careers depend on which hypotheses surviving, and what story about Alzheimer's disease the dominant institutions need to tell.

Monexus covered this story because the amyloid controversy is one of the clearest available case studies in how scientific paradigms interact with regulatory authority, commercial interest, and patient advocacy — and because the wire coverage of the Cochrane review treated it primarily as a biotech stock volatility story rather than as a fundamental epistemological challenge.