Anti-amyloid drugs in 2026: where lecanemab, donanemab and the rest stand

The drugs at a glance

Every figure in this table is read from this site's treatments file, which records the source for each one. "Pivotal n" is the number randomised in the main trial or trials.

The two approved drugs

Lecanemab (Leqembi)

Lecanemab binds soluble amyloid protofibrils, the intermediate clumps between single molecules and plaque. The FDA granted accelerated approval on 2023-01-06 and converted it to full approval that July after the phase 3 results. Its pivotal trial, Clarity AD (NCT03887455), randomised 1,795 people with early Alzheimer's and confirmed amyloid to fortnightly infusions or placebo for 18 months.

The primary outcome was change on the Clinical Dementia Rating Sum of Boxes. In the trial report in The New England Journal of Medicine, the mean change was 1.21 points with lecanemab and 1.66 with placebo, a difference of 0.45 on an 18-point scale, with a 95% confidence interval from 0.67 to 0.23. The treatments file records the same result as -0.45 point difference vs placebo, a 27% slowing of clinical decline (p<0.0001). The authors' own word for the benefit was "moderately", and they called for longer trials.

The label has changed since. The file records subcutaneous dosing with an autoinjector for maintenance, approved in 2025, and for starting treatment, approved in 2026, so the fortnightly infusion is no longer the only route. The European approval, in April 2025, is restricted to people who carry no copy, or only one copy, of the APOE4 variant, because of the ARIA risk described below.

Donanemab (Kisunla)

Donanemab targets a modified form of amyloid found in established plaques and clears existing plaque rather than preventing new aggregation. The FDA approved it on 2024-07-02. Its pivotal trial, TRAILBLAZER-ALZ 2 (NCT04437511), randomised 1,736 people, and its primary outcome was a different scale, the integrated Alzheimer's Disease Rating Scale, over 76 weeks.

The trial report in JAMA found a 35.1% slowing of decline in participants with low or medium tau at baseline and 22.3% in the whole trial population. The treatments file records the result as 35% slowing of decline vs placebo in low/medium tau population; 22% slowing in combined (overall) population. The gap between those two figures matters: the drug did more for people earlier in the tau process, which is the biological argument for treating early.

Two features set donanemab apart in practice. Dosing is monthly, and treatment can be stopped once scans show plaque has been cleared, so the course has an end. The dosing schedule was also changed after a follow-on trial, described next, cut the swelling risk.

ARIA: the risk that shapes everything

Amyloid-related imaging abnormalities are the harm that defines this class. ARIA-E is swelling or fluid in the brain; ARIA-H is small bleeds or surface iron deposits. Most episodes cause no symptoms and are found only on the scheduled MRI scans. A minority cause headache, confusion or seizures, and a small number have been fatal.

The numbers vary by drug, by dose and by genotype. For lecanemab, the treatments file records: Any ARIA (including asymptomatic radiographic events), Clarity AD: 21% (191/898) vs 9% (84/897) placebo; ARIA-E 13% (113/898) vs 2%; ARIA-H 17% (152/898) vs 9%. In the donanemab trial, the JAMA report found ARIA-E in 24.0% of treated participants against 2.1% on placebo, and ARIA-H in 31.4% against 13.6%, with three deaths in the donanemab group judged treatment-related. A secondary analysis of 3,030 participants across both donanemab trials, in JAMA Neurology, found that the risk of ARIA-E rose with the number of APOE4 alleles, with pre-existing microbleeds and surface iron deposits, with higher amyloid load and with higher blood pressure, and fell in people taking antihypertensives.

A 2025 meta-analysis in Neurology pooled nine phase 3 cohorts with 6,315 patients across the class. The pooled incidence of ARIA-E was 9.5% with a very wide confidence interval, 25.5% after removing outliers; symptomatic ARIA-E was 6.7%; ARIA-H was 17.8%. People with two copies of APOE4 had 5.6 times the odds of ARIA-E of non-carriers, and carriers of one copy 1.9 times. That is why the US labels require APOE testing before treatment and why Europe excludes homozygotes altogether.

The risk can be reduced. TRAILBLAZER-ALZ 6, reported in Alzheimer's & Dementia, randomised 843 people to the standard donanemab regimen or three alternatives. A gradual up-titration cut ARIA-E at 24 weeks from 23.7% to 13.7% with the same amyloid reduction, and the FDA approved that modified schedule in 2025. The treatments file now records donanemab's ARIA-E under the modified regimen as 16.0%.

Every label requires MRI before treatment and at fixed points during it. The lecanemab label specifies scans after one, two, three and six months of treatment; the donanemab label specifies scans before the second, third, fourth and seventh infusions. That monitoring is part of the cost, in money and in time, and it is part of what a family is signing up for.

The ones that failed, and why that matters

The approved drugs are the survivors of a long list. Four entries in the treatments file explain the shape of the field.

Aducanumab (Aduhelm) was approved on 2021-06-07 on the accelerated pathway, on the strength of amyloid reduction as a surrogate, after an advisory committee had voted against it. Its two identical phase 3 trials disagreed: EMERGE met its primary endpoint at the high dose and ENGAGE did not. The file records the safety result as: ARIA-E 35% (387/1105) at 10 mg/kg vs 3% placebo; ARIA-H 28% vs 9%; any ARIA including asymptomatic 41% vs 10%. The launch price of $56,000 a year was cut by about half within months. Medicare restricted coverage, uptake was poor, and the manufacturer discontinued the drug on 31 January 2024, citing business reprioritisation. Its accelerated approval was never converted.

Gantenerumab failed both of its phase 3 trials, GRADUATE I and II, in 1,965 people combined: relative reductions in decline of 8% and 6%, neither statistically significant, with pooled ARIA-E of 25%. The file notes that it removed roughly half the amyloid expected from earlier-phase data, which is widely taken as the reason it missed.

Solanezumab targeted single amyloid molecules rather than clumps, and failed three consecutive phase 3 trials, the last in 2,129 people. It then failed in two prevention trials as well. The contrast with lecanemab and donanemab, which target aggregated forms, is the clearest evidence that which form of amyloid a drug binds matters.

A 2022 review in Nature Reviews Drug Discovery draws the lesson: plaque has to be driven very low, to around 20 centiloids, before a clinical effect appears, and there is a lag between removal and benefit. On that reading the failures were failures of dose and timing. The alternative reading is in our article on what amyloid and tau do.

What is in phase 3

Remternetug, from the same company as donanemab, targets the same plaque form and is built for faster clearance and subcutaneous dosing. Its phase 3 programme includes a trial in early symptomatic disease with 1,667 enrolled, whose registered primary outcome is amyloid clearance on PET at 52 weeks rather than a clinical scale, a second trial in people with no or minimal impairment that measures time to clinical progression, and an arm in a prevention trial for inherited Alzheimer's. As of this site's last verification no phase 3 efficacy result had been posted.

Trontinemab is an anti-amyloid antibody fused to a fragment that carries it across the blood-brain barrier, so that far lower doses reach the brain. In early-phase data the file records that 91% of participants in the 3.6 mg/kg cohort became amyloid-negative on PET by 28 weeks, with a company-reported ARIA-E rate near 1% at the highest dose. Two phase 3 trials, with 2,161 enrolled between them, measure change in CDR-SB at 72 weeks and are due to complete in early 2028; a prevention trial in cognitively unimpaired people at risk is also registered. The file is explicit that the ARIA figure is early-phase and company-reported and should be treated as provisional until the phase 3 trials report.

Who is eligible, and what it costs

Both approved drugs are indicated for early Alzheimer's only, meaning mild cognitive impairment or mild dementia, and only with amyloid confirmed by PET or spinal fluid. In practice that excludes most people with a dementia diagnosis, who are past the mild stage by the time they are diagnosed. The trial populations also excluded many real-world patients, people on anticoagulants among them, and clinicians generally apply the same limits.

The eligibility question has an equity dimension the library documents. Across four early-Alzheimer's trials, Black, Hispanic and Asian participants in the US were two to four times more likely than white participants to be ruled ineligible on the amyloid biomarker criteria. Since the same biomarker threshold gates treatment, that gap carries over into who can be prescribed the drugs.

The treatments file records 2023 and 2024 list prices of $26,500 a year for lecanemab and $32,000 for a year of donanemab, before the cost of PET or spinal-fluid confirmation, APOE testing, the MRI schedule and the infusion visits. A 2024 cost-effectiveness model in Neurology concluded that lecanemab was not cost-effective against standard care at that price and would become so only below about $5,100 a year. Medicare covers the drugs under conditions; our getting covered page walks through the rules and how to appeal a denial.

Where this leaves a family in 2026

Our reading, which is interpretation and should be weighed as such. The approved drugs remove amyloid. The clinical benefit over 18 months is real, consistent across two independent programmes, and modest: a few tenths of a point on scales that run to 18 or more. The risk is substantial for APOE4 carriers and manageable, with monitoring, for most others. The cost in money, scans and appointments is high. Whether the trade is worth it depends on the person's stage, genotype, other illnesses, and what they would want a modest delay for.

The questions worth asking a clinician are concrete. Is this early-stage disease? Has amyloid been confirmed? What is the APOE genotype? What is on the baseline MRI? What would the monitoring schedule look like, and who pays for it? Our lecanemab and donanemab page goes deeper into the trial data and the disagreement between reviews about how large the benefit is.