Approved Alzheimer’s Drugs Cause Brain Swelling in 21% of Patients. A Gene-Silencing Injection Posted 0% Across 30 Months.
Alnylam's mivelsiran silences the gene that produces amyloid precursor protein, cutting amyloid production by 90% with a single spinal injection lasting six months. Phase 1 data presented at AAIC 2026 show zero cases of the brain swelling that afflicts up to one in five patients on approved antibody drugs. A dosing-burden analysis reveals why the upstream approach reshapes treatment economics for 30 million Alzheimer's patients and 6 million people with Down syndrome who cannot safely take the existing drugs at all.
Zero. Not "low incidence," not "comparable to placebo." Zero amyloid-related imaging abnormality events across 30 months of dosing, in a disease where the two FDA-approved drugs trigger brain swelling or microhemorrhages in 12.6% to 40% of patients. That number dropped at the Alzheimer's Association International Conference on Tuesday in a poster session about Alnylam Pharmaceuticals' mivelsiran, and it deserves to be read alongside every headline celebrating the antibody drugs that currently dominate Alzheimer's treatment.
Mivelsiran is an RNA interference therapeutic, a small interfering RNA that silences the messenger RNA encoding amyloid precursor protein. One intrathecal injection delivers the payload to the central nervous system, hijacking the cell's own gene-silencing machinery to shut down APP production at the genetic source. In the highest-dose group of Alnylam's Phase 1 trial, a single injection slashed cerebrospinal fluid levels of soluble APPβ by 89.9% and Aβ42, the amyloid fragment that aggregates into plaques, by 70.2%, with effects persisting for months and no serious adverse events attributed to the drug, no neuroinflammation, and no cerebrospinal fluid abnormalities. Patients in the multiple-ascending-dose arm received 50 milligrams every six months, reporting nothing worse than procedural headaches from the spinal tap.
Now compare. Lecanemab, marketed as Leqembi, requires biweekly intravenous infusions at $26,500 per year, slows cognitive decline by 27% on the CDR-SB scale, and produces ARIA in 21% of patients, with several trial deaths linked to brain hemorrhages. Donanemab, sold as Kisunla at $32,000 annually, triggers ARIA in 20% to 40% of recipients. Both drugs work downstream: they bind plaques already deposited in the brain and recruit immune cells to clear them, which inflames cerebral blood vessels weakened by decades of amyloid buildup.
Mivelsiran works upstream. It prevents the protein from being made. No excess protein means no new plaque accumulation and no inflammatory clearance reaction. Faucet versus mop.
Dosing Burden: 2,600 Visits Versus 200
Take 100 patients treated for a year. Lecanemab's protocol generates 2,600 infusion-center visits annually, each lasting roughly an hour with post-infusion observation, plus 400 mandatory quarterly MRI scans to watch for ARIA. When 21 patients develop brain swelling, the system absorbs emergency imaging, potential hospitalizations, treatment pauses, and agonizing decisions about whether to resume. Published ICER analyses put lecanemab's all-in cost near $35,000 per patient per year once monitoring and ARIA management are factored in.
Mivelsiran at two intrathecal injections annually generates just 200 procedures for the same cohort, and because zero patients develop ARIA under the current data, the mandatory MRI surveillance and emergency hospitalization pipeline that consumes so much of the antibody treatment infrastructure simply disappears. Lumbar punctures carry procedural risks, but the arithmetic is blunt: a 13-fold reduction in healthcare system contact.
Alnylam has not disclosed pricing, though its existing RNAi drug Amvuttra, used for transthyretin amyloidosis, lists at $119,351 per injection or $477,404 annually, which makes two mivelsiran doses at that tier roughly $240,000 per patient per year, nine times lecanemab's drug cost and a multiple that will survive contact with payers only if Phase 3 delivers cognitive benefit to match the biomarker improvements.
Six Million People Locked Out
For one population, pricing is secondary. Six million people worldwide have Down syndrome, caused by an extra copy of chromosome 21, which carries the APP gene, and every one of them overproduces amyloid precursor protein from birth, developing full Alzheimer's neuropathology by 40, clinical dementia in more than 90% of cases, and dying from its complications at rates that make Alzheimer's their leading cause of death after 35. It kills 70% of them. Life expectancy stalls at roughly 60, two decades short of the general population, and the gap is almost entirely explained by a single genetic overdose of one protein.
Lecanemab's appropriate-use guidelines explicitly advise against treating people with Down syndrome, because their lifelong saturation of amyloid deposits in cerebral blood vessel walls means the inflammatory clearance reaction triggered by antibody drugs carries an unacceptably high risk of the very brain swelling that already makes these drugs dangerous in patients with far less amyloid burden. In practice, the world's largest genetically determined Alzheimer's population cannot take the world's approved Alzheimer's drugs.
That is why the APPlauDS trial matters. Announced alongside the Phase 1 data, it is Alnylam's Phase 2 study of mivelsiran in Down syndrome-associated Alzheimer's across approximately 30 global sites. If the zero-ARIA profile holds, it would open the first therapeutic pathway this population has ever had.
Limitations
Phase 1 enrolled 20 patients. That is tiny. A zero in 20 carries a 95% confidence interval whose upper bound sits near 14%, which means the true ARIA rate could theoretically approach lecanemab territory. Only thousands of patients in Phase 3 will resolve that.
Mivelsiran prevents new amyloid but clears nothing already there. In patients with years of plaque accumulation, whether turning off the faucet alone slows cognitive decline remains unproven. If existing plaques drive ongoing damage, mivelsiran may need pairing with an antibody clearance agent, compounding costs rather than replacing them. And the deepest caution: a 90% biomarker reduction is not a cognitive benefit. History is strewn with Alzheimer's drugs that moved every biomarker beautifully and did nothing for the patients attached to them.
The Bottom Line
Every approved Alzheimer's therapy works downstream, clearing plaques and managing the inflammatory damage that clearance causes. Mivelsiran silences the gene producing the problem protein before it ever aggregates. Zero ARIA in 30 months is a data point, not a verdict, and 20 patients are not 2,000. But for 6 million people with Down syndrome, whose lecanemab guidelines begin with "do not treat," an upstream gene-silencing approach that avoids cerebral inflammation is not an incremental improvement. It is the first treatment option they have ever been offered.
What You Can Do
If you care for someone with Down syndrome, ask their neurologist about the APPlauDS trial on ClinicalTrials.gov under mivelsiran; Alnylam is recruiting globally. If you follow Alzheimer's therapeutics, track cAPPricorn-1 results in cerebral amyloid angiopathy expected in 2028, because that trial's primary endpoint, new lobar microbleeds by MRI, will test whether silencing APP translates into vascular protection. If you are a clinician weighing antibody drugs for a patient whose baseline MRI shows significant amyloid angiopathy, this safety data is worth reading before that conversation.