When the Medicine Comes With a Warning: Understanding ARIA, Leqembi, and Kisunla ... and What It Means for You

Something happened in my resident neurology clinic recently that I have been thinking about ever since.

A patient came in to discuss Leqembi and Kisunla — the two newer anti-amyloid medications that represent the first drugs in history to actually slow the progression of Alzheimer's disease rather than just manage its symptoms. The conversation was going well. Then I looked at the genetic results in the chart.

This patient carried two copies of the APOE-E4 gene — what we call a homozygote. And I had to stop and think carefully, because what I knew intuitively about the risk did not fully match what I found when I dug into the regulatory details. The FDA has not contraindicated these medications in APOE-E4 homozygotes. Europe has taken a different position, excluding this group from the approved indication entirely. The gap between those two stances is exactly the kind of clinical gray zone that deserves a careful, honest conversation — with patients, and with the broader public following this space.

This post is that conversation.

What Leqembi and Kisunla Actually Do

To understand ARIA, you first need to understand what these medications are doing inside the brain.

Alzheimer's disease is characterized by the accumulation of a sticky, toxic protein called amyloid-beta, which builds up in plaques between neurons over fifteen to twenty years before symptoms even appear. Leqembi (lecanemab) and Kisunla (donanemab) are monoclonal antibodies — precision immune proteins — that attach to this amyloid and flag it for removal by the brain's immune system. In the Phase 3 clinical trials, both drugs cleared amyloid substantially and slowed cognitive and functional decline by approximately 35% compared to placebo.

That is a genuine and meaningful advance. For the first time, we have medications that modify the underlying disease rather than merely masking its symptoms.

But amyloid removal is not a simple or entirely clean process. The brain's blood vessels are lined with amyloid deposits as well, and when the immune system begins aggressively clearing plaques, the vessels can become temporarily leaky and irritated. That leakiness is what produces ARIA.

What ARIA Actually Is

ARIA stands for Amyloid-Related Imaging Abnormalities. The name is clinical and dry, which somewhat obscures what it actually represents.

There are (at least) two types. ARIA-E (“E” refers to edema) — essentially, small areas of fluid accumulation or swelling in the brain tissue or surrounding spaces, visible on MRI. Think of it as temporary inflammation in the wake of amyloid clearance. ARIA-H (“H” refers to hemosiderin deposition) — essentially tiny micro-hemorrhages, small spots of blood that have leaked from vessels and left behind an iron-containing residue, also detectable on MRI.

Most ARIA is silent. Patients experience no symptoms whatsoever, and the abnormalities are only detected because of the scheduled monitoring MRI scans that are required during treatment. In the CLARITY-AD Phase 3 trial of Leqembi, ARIA-E occurred in about 13% of treated patients and ARIA-H in about 17% — but the majority of those cases were radiographic findings only, without any clinical consequence.

A smaller proportion of patients do develop symptoms. When symptomatic ARIA-E occurs, it can produce headaches, confusion, dizziness, visual disturbances, nausea, or in more serious cases, seizures or stroke-like episodes. Serious ARIA events — those requiring hospitalization or causing lasting neurological change — occurred in roughly 3% of APOE-E4 homozygotes in the clinical trials, compared to approximately 1% in other genetic groups.

The critical point is that ARIA is not permanent brain damage in most cases. When detected early on monitoring MRI, treatment is typically paused, the ARIA resolves, and treatment can often resume. The danger is in ARIA that goes undetected and is allowed to progress — which is precisely why the monitoring protocol exists.

Why APOE-E4 Changes Everything

APOE-E4 is the strongest known genetic risk factor for late-onset Alzheimer's disease. Most people carry two copies of the APOE-E3 variant — the neutral version. About 25% of the population carries one copy of APOE-E4, and roughly 2% to 3% carry two copies of APOE-E4 — what we call homozygotes.

Carrying one copy of APOE-E4 roughly triples the lifetime risk of Alzheimer's. Carrying two copies increases it by somewhere between eight and twelve times. The gene is not a guarantee — it is a risk amplifier.

But APOE-E4 does something else that is directly relevant here. It promotes the deposition of amyloid not just in brain tissue but specifically in the walls of blood vessels — a condition called cerebral amyloid angiopathy (CAA). Vessels laden with amyloid are more fragile, more prone to leaking, and more vulnerable to the disruption that occurs when amyloid clearance is initiated by these medications.

The numbers from the clinical trials make this concrete. In the CLARITY-AD trial of Leqembi, symptomatic ARIA-E occurred in approximately 9.2% of APOE-E4 homozygotes, compared to 1.7% in patients with one copy of the gene and 1.4% in those with no copies. ARIA-H — the micro-hemorrhage type — occurred in 39% of homozygotes, compared to 14% in heterozygotes and 12% in non-carriers. Serious ARIA events were three times more common in homozygotes than in the other groups.

The Regulatory Divide

Here is the part that I found genuinely surprising when I researched it for my patient.

The FDA, when approving both Leqembi and Kisunla, did not contraindicate their use in APOE-E4 homozygotes. The prescribing information carries strong warnings about the elevated risk in this group, and it recommends that APOE genotyping be performed before starting treatment so that the risk discussion can be properly informed. But the decision of whether to treat is left to the clinician and patient, not categorically prohibited.

Europe took a harder line. When the European Medicines Agency approved Leqembi in 2025, it specifically excluded APOE-E4 homozygotes from the approved indication entirely, citing the unfavorable benefit-risk profile in that population.

That transatlantic difference (which I have seen frequently in my career, including mostly prior to becoming a physician when I worked in the medical device industry) reflects a genuine scientific and ethical tension. A 2025 Bayesian reanalysis published in Alzheimer's & Dementia: Translational Research & Clinical Interventions examined the efficacy data for both drugs specifically in APOE-E4 homozygotes and found that the evidence of clinical benefit in this group was weak — the odds of no meaningful treatment effect were nearly three to four times greater than the odds of benefit. In other words, homozygotes appear to bear significantly more risk and may receive significantly less benefit than the broader population studied in the trials.

This does not mean that treatment is never appropriate for a homozygote. It means that the conversation with such a patient must be exceptionally careful, thorough, and individualized.

How ARIA Is Monitored During Year One

For patients who proceed with either Leqembi or Kisunla, the monitoring protocol during the first year is rigorous — and for good reason, since ARIA occurs most commonly in the first three to six months of treatment.

Before treatment begins, a baseline brain MRI is required. This establishes whether any pre-existing microhemorrhages or other vascular abnormalities are present that might increase risk further.

For Leqembi, the current FDA-recommended schedule requires MRI scans before the fifth infusion, before the seventh infusion, and before the fourteenth infusion — with an additional scan at one year for APOE-E4 carriers and for anyone who experienced any ARIA during the first year. Kisunla follows a similar schedule with scans at baseline, week four, week twelve, week twenty-four, and week fifty-two.

Patients and their families are instructed to report symptoms immediately — headache, confusion, dizziness, visual changes, nausea, or any new neurological symptom — because symptomatic ARIA requires prompt MRI evaluation and typically means treatment is paused until imaging confirms resolution.

CT scanning is not adequate for ARIA detection. Brain MRI with specific sequences — called gradient echo or susceptibility weighted imaging (SWI) — is required. This is not a minor logistical point. It means that patients pursuing these therapies need to have reliable access to MRI throughout their first year of treatment.

What This Means for Patients and Families

If you or a loved one is considering Leqembi or Kisunla, several practical points deserve emphasis.

APOE genotyping is not optional — it is a necessary part of the informed consent process for these medications. Knowing your genotype does not automatically determine whether you should or should not proceed, but it determines the level of risk you are accepting and the intensity of monitoring you will require.

ARIA is common but usually manageable when detected early. The monitoring protocol exists precisely to catch it before it becomes dangerous. Adherence to the MRI schedule should not be considered an inconvenience or an option — it is an integral part of the therapy itself.

For APOE-E4 homozygotes, this is a genuinely difficult conversation that deserves time and honesty. The elevated risk is real. The evidence of benefit in this specific genetic group is uncertain. The regulatory bodies of two major medical jurisdictions have reached different conclusions. That is not a reason to panic — it is a reason to engage carefully with a neurology healthcare provider who is current on the literature and who will take the time to work through the individual picture with you.

Takeaways

ARIA is the primary safety concern with anti-amyloid therapies. It reflects temporary vascular disruption during amyloid clearance and comes in two forms — swelling (ARIA-E) and micro-hemorrhage (ARIA-H). Most cases are silent and detected only on scheduled MRI.

APOE-E4 homozygotes carry substantially higher risk. Symptomatic ARIA-E occurred in over 9% and ARIA-H in 39% of homozygotes in the Leqembi trial, compared to roughly 1-2% and 12% respectively in non-carriers.

The FDA allows treatment in homozygotes with strong warnings. Europe does not. This regulatory divide reflects genuine uncertainty about whether the benefit outweighs the risk in this genetic subgroup.

Year-one MRI monitoring is not optional. A structured schedule of brain MRIs — before specific infusions and at one year — is required to detect ARIA early, when it is most safely managed.

APOE genotyping before starting treatment is essential. It does not make the decision for you, but it shapes the entire risk conversation that must precede it.

Citations and References

Cummings J, et al. Lecanemab: Appropriate Use Recommendations. Journal of Prevention of Alzheimer's Disease. 2023. https://pmc.ncbi.nlm.nih.gov/articles/PMC10313141/

NCBI Bookshelf. Lecanemab Therapy and APOE Genotype. Medical Genetics Summaries. 2024. https://www.ncbi.nlm.nih.gov/books/NBK605938/

Teipel S, et al. Clinical efficacy of anti-amyloid antibodies in apolipoprotein E ε4 homozygotes: A Bayesian reanalysis. Alzheimer's & Dementia: Translational Research & Clinical Interventions. 2025. https://pmc.ncbi.nlm.nih.gov/articles/PMC11982174/

Eisai/Biogen. LEQEMBI Prescribing Information and ARIA Safety Data. https://www.leqembihcp.com/about/safety

van Dyck CH, et al. Lecanemab in Early Alzheimer's Disease (CLARITY-AD). New England Journal of Medicine. 2023. https://doi.org/10.1056/NEJMoa2212948