Medicines for Alzheimer’s

No medicine cures Alzheimer’s. Some ease symptoms for a while. Two newer drugs slow measured decline a little in the early stage, at real cost and real risk. Here is what each one does, in plain numbers, with the name of whoever paid for the research.

The short version

Two kinds of Alzheimer’s medicine are approved in the United States.

• Symptom drugs. Donepezil, rivastigmine, galantamine, memantine and combinations of them. They can help memory and thinking a little for a while.
• Anti-amyloid antibodies. Lecanemab (brand name Leqembi) and donanemab (Kisunla). They clear amyloid plaque from the brain and slow measured decline modestly. They are given by drip or injection and need regular MRI safety scans.

A third group treats single symptoms, such as agitation or poor sleep.

Experts disagree about whether the antibody benefit is big enough for a family to notice. We set out both sides further down this page. Whether any of these is right for your parent is a decision for their doctor, with you in the room.

The older symptom drugs

Three of these drugs are cholinesterase inhibitors. They slow the breakdown of acetylcholine, a chemical messenger that nerve cells use for memory. Memantine works differently. It dampens excess signalling by another messenger, glutamate.

How much they help

The best summary we could open is a Cochrane review of the three cholinesterase inhibitors. Cochrane is a network that pools all the trials on a question. On a 70-point thinking test, people on the drugs scored about 2.4 points better than people on placebo. That is a small change. Some families notice a difference and many do not. This is our reading of the numbers.

Solid4/5Funder not confirmed

Donepezil, rivastigmine and galantamine: the Cochrane review

Birks J, Cochrane Database of Systematic Reviews 2006 (CD005593) Read the study

Design

Systematic review of 10 randomized trials

People

10 trials pooled

Length

Not reported in the summary we read

Result On a 70-point thinking test (ADAS-Cog), people on the drugs scored on average 2.37 points better than people on placebo. More people stopped the drugs than stopped placebo (29 in 100 vs 18 in 100). Nausea, vomiting and diarrhoea were more common on the drugs. The three drugs worked about equally well; donepezil had fewer side effects than rivastigmine. Cochrane now marks this review as “considerably out of date.”

Who paid: Not reported (Cochrane review). The original drug trials were largely run by the makers; not yet confirmed trial by trial.

Conflicts: Not reported

Cochrane itself says this review is “considerably out of date.” Newer Cochrane reviews of donepezil (2018) and memantine (2019) exist. We could not open them, so we do not quote their numbers here.

Side effects

For the cholinesterase inhibitors, the common ones are nausea, diarrhoea, poor appetite and vivid dreams. In the Cochrane review, 29 in 100 people stopped the drug, compared with 18 in 100 on placebo. A slow heart rate and fainting are also often listed, though we could not confirm the exact label wording. For memantine, dizziness, headache and confusion are commonly reported; we have not confirmed these against the label either. Ask the pharmacist to go through the full list with you.

The anti-amyloid antibodies

These are lab-made antibodies that latch onto amyloid, the sticky protein that builds up as plaque in Alzheimer’s, and help the immune system clear it. Both clearly remove plaque. The question is how much that helps the person.

Lecanemab (Leqembi and Leqembi Iqlik)

• Who makes it: Eisai and Biogen. The antibody was originally developed by BioArctic.
• How it is given: a drip (IV infusion) every 2 weeks for 18 months, then a maintenance phase.
• Approved: accelerated approval January 2023, full approval July 2023.
• At-home injections: Leqembi Iqlik is an autoinjector pen given under the skin once a week. It was approved for maintenance in August 2025. On 13 July 2026 the FDA approved it for starting treatment too (500 mg weekly, as two 250 mg injections), which it called “the first at-home starting dosage regimen.” Patients can switch between the drip and the pen.
• How the pen was approved: it produces the same drug levels in the blood as the drip. It was not tested in its own large trial of benefit.

In the main trial, Clarity AD, decline on the 18-point CDR-SB scale was 27% slower. In absolute terms that is 0.45 points out of 18 over 18 months.

Promising4/5Industry

Clarity AD: lecanemab (Leqembi) in early Alzheimer’s

van Dyck CH et al., New England Journal of Medicine 2023 Read the study

Design

Randomized, placebo-controlled trial

People

1,795 people with mild cognitive impairment or mild dementia and confirmed amyloid

Length

18 months

Result On the 18-point CDR-SB dementia scale, people on lecanemab worsened by 1.21 points and people on placebo by 1.66 points. The difference was 0.45 points, reported as “27% slower decline.” Brain swelling on scans (ARIA-E) showed up in 12.6% on the drug vs 1.7% on placebo, usually without symptoms. Infusion reactions in 26.4%.

Who paid: Eisai and Biogen, which sell the drug

Conflicts: Company-funded trial of its own product

Donanemab (Kisunla)

• Who makes it: Eli Lilly.
• How it is given: a drip once a month. Treatment can stop once the plaque has been cleared to low levels.
• Approved: FDA 2 July 2024; UK October 2024; EU September 2025, after an initial negative opinion.
• Slower start, less swelling: in July 2025 the FDA approved a gentler starting schedule. In Lilly’s TRAILBLAZER-ALZ 6 trial, brain swelling at 24 weeks fell from 24 in 100 people to 14 in 100, with similar plaque removal (company figures).

In TRAILBLAZER-ALZ 2, decline was 22% slower on Lilly’s main scale for the whole group, and 35% slower in people with less tau. On the 18-point CDR-SB scale the difference was 0.70 points for the whole group.

Promising4/5Industry

TRAILBLAZER-ALZ 2: donanemab (Kisunla) in early Alzheimer’s

Sims JR et al., JAMA 2023 Read the study

Design

Randomized, placebo-controlled trial

People

1,736 people with mild cognitive impairment or mild dementia and confirmed amyloid

Length

76 weeks (about 18 months)

Result On the 18-point CDR-SB scale, the whole group worsened 1.72 points on donanemab vs 2.42 on placebo, a 0.70-point difference. In people with low or medium tau the difference was 0.67 points (1.20 vs 1.88). Brain swelling (ARIA-E) in 24.0% vs 2.1% on placebo; any ARIA in 36.8% vs 14.9%. Deaths: 16 on the drug (3 judged treatment-related) vs 10 on placebo (1 treatment-related).

Who paid: Eli Lilly, which sells the drug

Conflicts: Run by Lilly; authors affiliated with Lilly

What “half a point” means

The CDR-SB rates six areas of life, including memory, judgement, hobbies and personal care, each from 0 to 3. A 0.45-point difference is less than one step of “questionable” impairment in one of those six areas. One way to picture it: after 18 months, people on the drug were roughly where people on placebo had been 4 to 5 months earlier. That comparison is our estimate from the trial numbers, not a figure the trials reported.

The risks in numbers

• Lecanemab: brain swelling on scans in 12.6 in 100 people (1.7 in 100 on placebo). Symptoms from ARIA in about 3 in 100; serious symptoms in 0.7 in 100. A brain bleed larger than 1 cm in 0.7 in 100, vs 0.1 in 100 on placebo. Infusion reactions in about 26 in 100.
• Donanemab: brain swelling in 24 in 100 (2.1 in 100 on placebo). Any ARIA, swelling or small bleeds, in 36.8 in 100 (14.9 in 100 on placebo). 16 people died on the drug and 10 on placebo; 3 deaths on the drug and 1 on placebo were judged treatment-related.
• Both: the brain shrank slightly faster on the drug than on placebo. What that means is not yet known.

Outside the US, regulators have been more cautious. The EU and UK restrict use by APOE4 status; the EU licence for donanemab excludes people with two copies. England’s NICE declined to fund lecanemab on the NHS, saying the small benefit does not justify the cost. Australia’s regulator declined to register it in 2025.

Who qualifies

• Mild cognitive impairment or mild dementia due to Alzheimer’s. Not moderate or severe dementia.
• Amyloid confirmed by a PET scan or a spinal fluid test.
• Able to have regular MRI safety scans.
• Blood thinners and two copies of APOE4 raise the risk and should be discussed with the doctor.

What it costs

• Lecanemab: list price about $26,500 a year for the drip. That excludes MRI and PET scans, infusion visits and genetic testing. The price of the Iqlik pen was not given when it was approved.
• Donanemab: about $32,000 for a year. Because treatment can stop once plaque clears, the total depends on how long it runs: roughly $12,522 for 6 months to $48,896 for 18 months.

Medicare and the registry

Medicare covers both drugs under a rule called Coverage with Evidence Development. The prescribing clinician must enter your parent in the CMS National Patient Registry, or an approved study, at the start of treatment. They then report back every 6 months for up to 24 months. The underlying decision dates from April 2022. The CMS page, last updated 10 March 2026, shows no recent change. You usually still owe the Part B share of the cost unless you have supplemental coverage. Many private insurers add their own limits. The Alzheimer’s Association has asked CMS to drop the registry requirement since 2022. Read the rule on the CMS coverage page.

Drugs for specific symptoms

These do not treat memory. They target one hard symptom. For non-drug ways to handle agitation and sleep problems, see our page on living with Alzheimer’s.

Agitation: brexpiprazole (Rexulti)

Approved in May 2023 as the first drug for agitation due to Alzheimer’s dementia. It is an antipsychotic, so it carries the boxed warning that older people with dementia on these drugs have a higher risk of death. Common side effects: headache, dizziness, urinary infection and sleep problems.

Promising3/5Industry

Brexpiprazole (Rexulti) for agitation in Alzheimer’s dementia

US Food and Drug Administration approval announcement, 12 May 2023 Read the study

Design

Two placebo-controlled trials

People

Not reported in the sources we read

Length

12 weeks each

Result The 2–3 mg dose lowered agitation scores (Cohen-Mansfield Agitation Inventory) more than placebo. An FDA advisory committee voted 9–1 in favour. It carries a boxed warning: older people with dementia who take antipsychotics have a higher risk of death. One secondary source reports deaths of 0.9% on the drug vs 0.3% on placebo in the trials; we have not confirmed that against the label.

Who paid: Otsuka and Lundbeck, which sell the drug

Conflicts: Maker-sponsored approval trials

Agitation: dextromethorphan–bupropion (Auvelity), new in 2026

Approved on 30 April 2026. The FDA calls it “the first non-antipsychotic drug” for agitation in Alzheimer’s dementia, so it does not carry the antipsychotic death warning. It does carry warnings for seizures, raised blood pressure, and the antidepressant warning about suicidal thoughts in young people. Some experts have questioned how well it works.

Promising3/5Industry

Dextromethorphan–bupropion (Auvelity) for agitation in Alzheimer’s dementia

US Food and Drug Administration approval announcement, 30 April 2026 Read the study

Design

A 5-week placebo-controlled trial (ADVANCE-1) and a relapse-prevention trial (ACCORD-2)

People

Not reported in the sources we read

Length

5 weeks (ADVANCE-1)

Result In ADVANCE-1, agitation scores improved significantly more than on placebo. In ACCORD-2, 8.4% of people kept on the drug relapsed vs 28.6% switched to placebo. A second short trial (ADVANCE-2) is reported to have missed its main goal, but we have not confirmed that. Side effects included dizziness, nausea, headache, diarrhoea, sleepiness and dry mouth.

Who paid: Axsome Therapeutics, which sells the drug

Conflicts: Maker-sponsored approval trials

Sleep: suvorexant (Belsomra)

Its label was expanded in February 2020 to cover insomnia in mild to moderate Alzheimer’s. It added about 28 minutes of sleep a night compared with placebo. Watch for next-day drowsiness and falls, and it comes with a driving caution.

Promising3/5Industry

Suvorexant (Belsomra) for poor sleep in Alzheimer’s

Merck announcement of Phase 3 results (company claim); FDA label expanded 3 February 2020 Read the study

Design

Randomized, placebo-controlled trial

People

285 people with mild to moderate Alzheimer’s and insomnia

Length

4 weeks

Result Total sleep time rose by 73 minutes on the drug vs 45 minutes on placebo, a gain of 28 minutes a night. Side effects included sleepiness (4%), dry mouth (2%) and falls (2%), plus next-day drowsiness.

Who paid: Merck, which sells the drug

Conflicts: Maker-sponsored trial

All of them side by side

“Not confirmed” means we could not open a primary source for that figure during our review. Prices are US list prices and exclude scans and visits.

Is the benefit big enough to notice?

This is the most argued-over question in Alzheimer’s medicine today. Both sides include serious scientists.

What the trials measured

Lecanemab: 0.45 points on the 18-point CDR-SB over 18 months. Donanemab: 0.67 to 0.70 points. Researchers have tried to define the smallest change a person would notice. One group (Andrews and colleagues, 2019) put it at 1 to 2 points. Another, from the drug company Roche (Lansdall and colleagues, 2022), put it at 1.0 to 2.5 points. Both groups said these numbers apply to one person’s change over time, not to the gap between two group averages, and that the bar is lower in early disease.

The case that it matters

• A 25–35% slowing adds up. If the gap kept widening at the same rate, 0.45 points could reach about 1 point by year 3.
• Lilly and Eisai report that the benefit keeps growing in the follow-on phases where everyone gets the drug (company claims, without a placebo group).

The case that it doesn’t

• 0.45 points is less than one “questionable” rating in one of six areas of life.
• Nobody has shown the gap really keeps widening in a straight line.
• Robert Howard of University College London points out the benefit is about 2% of the scale and reached statistical significance only because the trials were very large.
• The brain shrinks faster on the drugs, and the swelling and bleeding risks are real.

The April 2026 Cochrane review, and the response

In April 2026, Cochrane published a review of 17 trials in 20,342 people. It concluded the benefit was “absent or trivial” and that the drugs “likely increase the risk of swelling and bleeding in the brain.”

Didn’t work4/5Funder not confirmed

Cochrane review of all anti-amyloid antibodies

Nonino F, Richard E et al., Cochrane Database of Systematic Reviews 2026 (CD016297) Read the study

Design

Systematic review of 17 randomized trials

People

20,342

Length

Trials of varying length

Result Concluded that effects on thinking and dementia severity were “absent or trivial, falling well below established thresholds” for a difference people would notice, and that the drugs “likely increase the risk of swelling and bleeding in the brain.” Critics point out that 15 of the 17 trials tested antibodies that failed or were withdrawn, and only 2 tested the approved drugs. We could not open the full text, so its pooled numbers are not shown here.

Who paid: Not yet confirmed

Conflicts: Not yet confirmed

Many Alzheimer’s researchers objected at once, including Bart De Strooper of the UK Dementia Research Institute, John Hardy, Jonathan Schott and Suzanne Schindler. Their main point: 15 of the 17 trials tested antibodies that failed or were withdrawn, and only 2 tested the drugs now in use. De Strooper said pooling them “turns therapeutic progress into statistical noise.” International clinicians published a formal rebuttal, “Recent Cochrane review has serious flaws,” in Alzheimer’s & Dementia (Snyder and colleagues, 2026). We have not read the rebuttal in full.

Our honest summary

The two approved antibodies clearly remove amyloid and modestly slow measured decline over 18 months. Whether that slowing is large enough for a family to notice is genuinely disputed. The benefit comes with a real risk of brain swelling or bleeding, which is usually silent but occasionally serious or fatal. Both trials were paid for by the companies that sell the drugs. The makers also ran most of the trials behind the other drugs here, though we have not confirmed that drug by drug. The antibody trials were large and published in major journals, and the risk numbers above come from those same company trials. To see how drug funding compares with supplement and program funding, read our follow-the-money page. For what is coming next, see our clinical trials page.

Questions to take to the doctor

• Is my parent at a stage where any of these drugs could help?
• For the antibodies: has amyloid been confirmed? Should we test for APOE4? What about her blood thinner?
• How will we know if a symptom drug is working, and when should we stop it?
• Which symptoms of brain swelling should send us to the emergency room?
• What will we pay after Medicare, and who enters her in the registry?

Words used on this page

Amyloid

A sticky protein fragment that clumps into plaques between nerve cells in Alzheimer’s.

Cholinesterase inhibitor

A drug that slows the breakdown of acetylcholine, a chemical messenger used for memory. Donepezil, rivastigmine and galantamine.

CDR-SB

Clinical Dementia Rating, Sum of Boxes. A doctor rates six areas of daily life from 0 to 3, for a total of 0 to 18. Higher is worse.

ARIA

Amyloid-related imaging abnormalities. Swelling (ARIA-E) or small bleeds (ARIA-H) in the brain seen on MRI, caused by anti-amyloid drugs.

APOE4

A common version of a gene that raises Alzheimer’s risk. People with two copies (one from each parent) have the highest ARIA risk on antibody drugs.

Coverage with Evidence Development

A Medicare rule that pays for a treatment only if the patient’s results are collected in a registry or study.