Evidence
Follow the money
Who pays for Alzheimer’s research, how to read a funding statement, the approval that went wrong, the fraud cases, and why sellers on the alternative side deserve the same scrutiny as drug companies.
Distrust of drug-company research is earned. This page shows where that distrust is justified and where the evidence holds up anyway. It also applies the same test to supplement makers, program sellers, device companies and food industry groups, who pay for many of the “natural” studies.
Who pays for Alzheimer’s research
There are four kinds of funder. On this site each study carries a label for one of them.
- Public / nonprofit Governments and universities. In the US that mainly means the National Institutes of Health (NIH) and its National Institute on Aging (NIA).
- Charity Disease foundations and philanthropy, such as the Alzheimer’s Association.
- Industry A drug, device or food company testing its own product.
- Seller conflict The maker of a supplement or program, or an author who profits from it.
For drugs, industry dominates the late stage. The annual count of Alzheimer’s drug trials by Jeffrey Cummings and colleagues found that in 2026, industry sponsored 59% of all trials and 72% of Phase 3 trials, the large final-stage tests used for approval (2026 pipeline report). For the details, see our clinical trials page.
| Treatment | Who paid | People | What they found |
|---|---|---|---|
| LecanemabLeqembi | Industry Eisai, Biogen | 1,795 | 0.45 points slower decline on an 18-point scale; brain swelling in 12.6 in 100 |
| DonanemabKisunla | Industry Eli Lilly | 1,736 | 0.70 points slower on the same scale; swelling or bleeding in 36.8 in 100 |
| Semaglutide pill | Industry Novo Nordisk | 3,808 | No slowing; the company reported its own failure |
| Ginkgo biloba | Public NIH | 3,069 | No fewer dementia cases |
| Cocoa flavanols | Industry Mars Edge, with NIH | 2,262 | No benefit to thinking overall |
| Souvenaid drink | Industry Danone Nutricia (later years) | 311 | Null at the planned 24 months; positive later, with heavy dropout |
| Saffron vs donepezil | Seller Saffron company co-funded | 54 | Saffron about as good as the drug |
| Lion’s mane extract | Seller Grape King Bio | 41 | Some scores improved, others didn’t |
| ReCODE (Bredesen) | Seller Senior author (Bredesen) is Apollo Health’s chief science officer | 25 | Scores rose; no comparison group |
| Ornish program | Seller Donors; Ornish holds equity | 51 | Some improvement vs controls over 20 weeks |
| Aerobic exercise (EXERT) | Public NIA | 296 | No better than stretching; neither group declined |
People = the size of the main trial. Each study has a full card, with sources, on the page linked below.
A pattern worth knowing
Look across this whole site and a pattern shows up. The large trials paid for with public money, testing ginkgo, fish oil, vitamin D, the MIND diet and exercise, mostly found no effect. Many of the small positive studies of supplements were funded or supplied by the maker. That does not prove bias. But it is the pattern to watch for.
It also cuts the other way. Industry-funded trials sometimes report failure. Novo Nordisk announced its own semaglutide flop. The cocoa trial funded by Mars found no overall benefit.
Publicly funded trials with disappointing results
GEM: ginkgo biloba to prevent dementiaDeKosky ST et al., JAMA 2008 · 3,069 older adults · About 6 years
VITAL: fish oil and thinkingKang JH et al., Alzheimer’s & Dementia: TRCI 2022 · About 3,300 older adults · 2–3 years
VITAL: vitamin D3, 2,000 IU a dayKang JH et al., Scientific Reports 2021 · 4,218 older adults · 2–3 years
MIND diet, tested head-to-headBarnes LL et al., New England Journal of Medicine 2023 · 604 older adults · 3 years
EXERT: aerobic exercise vs stretching in mild memory lossBaker LD et al., Alzheimer’s & Dementia 2025 · 296 people with mild cognitive impairment · 18 months
VALAD: the herpes antiviral valacyclovir in early Alzheimer’sDevanand DP et al., JAMA 2025 · 120 people with early Alzheimer’s who carried herpes simplex virus · 78 weeks
Company-funded drug trials
Clarity AD: lecanemab (Leqembi) in early Alzheimer’svan Dyck CH et al., New England Journal of Medicine 2023 · 1,795 people with mild cognitive impairment or mild dementia and confirmed amyloid · 18 months
TRAILBLAZER-ALZ 2: donanemab (Kisunla) in early Alzheimer’sSims JR et al., JAMA 2023 · 1,736 people with mild cognitive impairment or mild dementia and confirmed amyloid · 76 weeks (about 18 months)
evoke and evoke+: semaglutide (the Ozempic family) as a pillNovo Nordisk topline announcement, 24 November 2025 (company release) · 3,808 people with early Alzheimer’s · 2 years
How to read a funding and conflict statement
Almost every published paper has two short sections, usually near the end, called something like “Funding” and “Conflicts of interest” (or “Competing interests” or “Disclosures”).
Five questions for any study
For US doctors, the government’s Open Payments database shows what drug and device companies have paid them. It is the best single check before trusting a doctor’s recommendation. For a trial, the ClinicalTrials.gov listing names the sponsor.
The aducanumab approval
Aducanumab (brand name Aduhelm), made by Biogen, is the clearest recent case of approval running ahead of the evidence.
- March 2019: Biogen stopped its two identical trials, EMERGE and ENGAGE, because they looked unlikely to succeed.
- Later analysis: on the high dose, EMERGE showed 22% slower decline. ENGAGE showed 2%, which could easily be chance.
- November 2020: the FDA’s outside advisory committee voted against approval.
- 7 June 2021: the FDA granted accelerated approval anyway. Three advisers resigned.
- Price: $56,000 a year, later cut to $28,200.
- December 2022: a US House investigation found the FDA “broke its own protocols” in working with Biogen.
- January 2024: Biogen abandoned the drug.
This episode is one reason Medicare now pays for the newer antibody drugs only if patients are entered in a registry. The two drugs approved since, lecanemab and donanemab, went through full-size trials that met their main goals; see our medicines page for how big that benefit is.
Research fraud cases
The Lesné paper
In 2006, a paper in Nature led by Sylvain Lesné claimed that one specific form of amyloid, called Aβ*56, impaired memory. It was cited about 2,300 times.
- Matthew Schrag, a Vanderbilt neurologist, found apparent image manipulation in about 20 of Lesné’s papers and reported it to NIH and the journals in January 2022.
- Science published an investigation by journalist Charles Piller in July 2022.
- The University of Minnesota’s own review at that stage reported “no findings of research misconduct.”
- Nature retracted the paper on 24 June 2024, citing “excessive manipulation, including splicing, duplication and the use of an eraser tool.”
- Lesné resigned from the University of Minnesota, effective 1 March 2025. He had received more than $7 million in NIH funding.
The Masliah papers
In September 2024, Science reported suspected image manipulation in about 132 papers by Eliezer Masliah. At the time he directed the Division of Neuroscience at the National Institute on Aging. NIH announced a finding of research misconduct that month. The affected work included research behind prasinezumab, a Parkinson’s antibody, and cerebrolysin. By December 2025, 18 of the papers had been retracted.
What these cases do and don’t mean
These cases damaged trust and wasted years of work and money. They do not by themselves disprove the amyloid theory, which also rests on genetics and on many independent labs. It is fair to be angry. It is not accurate to say the whole field is fraudulent.
Seller conflicts on the alternative side
Readers who distrust drug companies are right to ask who paid. The same question applies to supplements, programs and clinics. Here are cases from our other pages, stated as fact.
- Apollo Health and ReCODE. Dr. Dale Bredesen is chief science officer of Apollo Health, which sells the ReCODE program ($75 a month), PreCODE ($39.95 a month), a blood test, and practitioner certification ($2,495). He is behind the main published studies of it. A 2020 critique in Lancet Neurology by UCSF neurologist Joanna Hellmuth argued these interests were not properly disclosed. See our alternative programs page.
- Ornish Lifestyle Medicine. Dr. Dean Ornish holds equity in the company that sells his program. His 2024 randomized trial was paid for by private donors.
- Grape King Bio and lion’s mane. The maker of the lion’s mane extract paid the salaries of five of the authors of a 2020 trial.
- Cognito Therapeutics and 40 Hz light and sound. The company ran the OVERTURE trial and the authors were company employees. The trial missed its main goal.
- Aviv and hyperbaric oxygen. The senior author of a hyperbaric oxygen study co-founded Aviv Scientific, which sells the treatment at Aviv Clinics.
- Danone and Souvenaid. In the LipiDiDiet trial of the Souvenaid drink, the planned 24-month result, funded by the EU, was null. Danone’s Nutricia division funded the later years, which reported benefits among far fewer remaining participants; the trial statistician was paid by Nutricia.
- Mars and cocoa. The COSMOS cocoa trials were funded by Mars Edge, part of Mars Inc. They found no overall benefit to thinking.
- The blueberry industry. Many blueberry and memory trials were funded or supplied by the US Highbush Blueberry Council or the Wild Blueberry Association of North America. In the US POINTER lifestyle trial, the Highbush Council gave the coached group monthly blueberry rebates.
- Saffron. A saffron company, Green Plants of Life (IMPIRAN), co-funded the best-known trial comparing saffron with donepezil and supplied the saffron.
None of this makes these results false. Several of these studies are careful, and some reported null results. It means they need independent repeats, just as a drug company’s trial does.
Studies with a seller or industry link
ReCODE precision-medicine pilotToups K et al., Journal of Alzheimer’s Disease 2022;88(4):1411–1421 · 25 with mild cognitive impairment or early Alzheimer’s · 9 months
“Reversal of cognitive decline”: the first Bredesen case seriesBredesen DE, Aging 2014 · 10 · Not reported
Ornish intensive lifestyle program in early Alzheimer’sOrnish D et al., Alzheimer’s Research & Therapy 2024 · 51 enrolled, 49 completed; mild cognitive impairment or early Alzheimer’s · 20 weeks
Lion’s mane extract in mild Alzheimer’sLi IC et al., Frontiers in Aging Neuroscience 2020 · 49 people with mild Alzheimer’s enrolled (41 in the main analysis) · 49 weeks
OVERTURE: 40 Hz flickering light and soundHajós M et al., Frontiers in Neurology 2024 · 76 people with mild to moderate Alzheimer’s · 6 months
Hyperbaric oxygen in older adults and Alzheimer’s miceShapira R et al., Aging 2021 · Small number of patients (no control group), plus mice · Not reported
Saffron compared with donepezil (Aricept)Akhondzadeh S et al., Psychopharmacology 2010 · 54 people with mild to moderate Alzheimer’s · 22 weeks
ACTIVE: speed-of-processing training, 20 years onCoe et al., Alzheimer’s & Dementia: TRCI 2026 · 2,021 older adults · 20 years
Souvenaid (Fortasyn Connect) in very early Alzheimer’s: LipiDiDietSoininen H et al., Alzheimer’s & Dementia 2021 · 311 with prodromal Alzheimer’s (only 81 in the 36-month analysis) · 36 months
COSMOS: cocoa flavanolsBaker LD et al., Alzheimer’s & Dementia 2022; Vyas CM et al., AJCN 2024 · 2,262 · 2–3 years
COSMOS: a daily multivitaminBaker LD et al., Alzheimer’s & Dementia 2022; Yeung LK et al., AJCN 2023 · 2,262 and about 3,500 · 2–3 years
US POINTER: coached lifestyle program vs do-it-yourselfBaker LD et al., JAMA 2025 · 2,111 at-risk older adults · 2 years
The bottom line
Ask who paid, every time, whether the product is a $26,500 infusion or a bottle of capsules. Then look past the funder to the study itself: how many people, how long, compared with what, and whether the main goal was met. Every study on this site shows all of that, and you can filter by funder in the study database. How we grade studies is explained on our about page.