Sorry to be the one to tell you this, but your brain has been accumulating sticky protein clumps for possibly decades, and the newest drug designed to clean them up might cause its own brand of trouble - especially if you lost the genetic lottery in one very specific way.

Lecanemab (sold as Leqembi, because pharma loves a name that sounds like a wellness retreat) is the monoclonal antibody that got everyone excited when the FDA gave it full approval in July 2023. It targets amyloid beta protofibrils - basically the most toxic form of the gunk that builds up in Alzheimer's brains - and tells them to kindly vacate the premises. In the landmark CLARITY AD trial, it slowed cognitive decline by about 27% over 18 months compared to placebo (van Dyck et al., 2023). Not a cure, but for a disease with almost zero wins in the treatment column, even a modest slow-down felt like finding water in the desert.

The Side Effect Nobody Warned You About (Just Kidding, They Definitely Did)

Here's where things get interesting, and by "interesting" I mean "the kind of thing that keeps neurologists checking MRIs at 2 a.m." Enter ARIA - amyloid-related imaging abnormalities. It comes in two flavors: ARIA-E (brain swelling, the dramatic one) and ARIA-H (tiny brain bleeds, the sneaky one). Think of it as your brain's immune system getting a little too enthusiastic about the cleanup job, like a demolition crew that accidentally takes out a load-bearing wall.

A new retrospective study from Norton Neuroscience Institute tracked 187 patients who received lecanemab over two years in a real-world clinical setting - not the carefully controlled bubble of a clinical trial, but the messy reality of actual medicine (Shields et al., 2026). Out of 175 patients who got follow-up brain MRIs, 22.3% showed evidence of ARIA. That's roughly one in five patients whose brains decided to throw a small tantrum in response to treatment.

Your Genes Are Snitching on You

Now for the plot twist that geneticists saw coming from a mile away. The study found that your APOE gene - specifically, how many copies of the epsilon-4 (ε4) variant you carry - dramatically changes your ARIA risk. Patients who were ε4 homozygous (two copies of the risky variant, a genetic double whammy affecting about 15% of Alzheimer's patients) had a 48% rate of ARIA. Nearly half. Compare that to the overall 22.3% rate, and you start to see the problem.

The numbers get worse the more you dig. Homozygous ε4 carriers had statistically significant increases across the board: more ARIA-E (p = 0.041), more ARIA-H (p = 0.004), and more any-ARIA-at-all (p = 0.002). Five patients developed symptomatic ARIA severe enough to pause treatment, and three of them were ε4 homozygous. This tracks with earlier phase 2 data showing ε4 homozygotes hitting a roughly 50% ARIA rate (Honig et al., 2023).

And here's the kicker that should make everyone sit up a little straighter: a Bayesian reanalysis of the phase 3 trial data recently found that lecanemab may not even work for ε4 homozygotes. The probability of no treatment effect was three to five times higher than the probability of benefit (Teipel et al., 2025). So the patients most likely to get side effects might also be the ones least likely to benefit. (Science can be cruel like that.)

What This Means for Real Humans

The FDA already recommends APOE testing before starting lecanemab, and this study hammers home why that matters. It's not a contraindication - having ε4 doesn't automatically disqualify you - but it should absolutely change the conversation between doctors and patients. Think of it as the neurological equivalent of checking the weather before going skydiving.

Real-world data from other centers are broadly consistent. A prospective Chinese cohort found a 19% ARIA rate with all cases asymptomatic (Shang et al., 2025), and Eisai's own two-year U.S. data showed 87.4% of patients continuing treatment - suggesting most people tolerate the drug reasonably well. The Norton study's 22.3% ARIA rate actually falls right in line with trial data, which is reassuring in the "at least it's not worse than we expected" sense.

The Bottom Line (Without Using the Word "Groundbreaking")

Lecanemab works for some people, causes brain-visible side effects in about a fifth of patients, and carries particularly thorny risk-benefit math for anyone carrying two copies of APOE ε4. The science here isn't telling us to abandon ship - it's telling us to navigate more carefully, with better genetic maps and more frequent MRI check-ins. Your brain's cleanup crew is on the job. We just need to make sure they don't trash the place in the process.

References

1. Shields LBE, Hust HS, Cooper GE, et al. Lecanemab over a two-year duration: Key insights from a regional specialty medical center. The Journal of Prevention of Alzheimer's Disease. 2026. DOI: 10.1016/j.tjpad.2026.100489. PMID: 41581261.

2. van Dyck CH, Swanson CJ, Aisen P, et al. Lecanemab in Early Alzheimer's Disease. New England Journal of Medicine. 2023;388(1):9-21. DOI: 10.1056/NEJMoa2212948. PMID: 36449413.

3. Honig LS, Barakos J, Dhadda S, et al. ARIA in patients treated with lecanemab (BAN2401) in a phase 2 study in early Alzheimer's disease. Alzheimer's & Dementia: Translational Research & Clinical Interventions. 2023;9(1):e12377. DOI: 10.1002/trc2.12377. PMID: 36949897. PMCID: PMC10026083.

4. Teipel S, Tang Y, Khachaturian A. Clinical efficacy of anti-amyloid antibodies in apolipoprotein E ε4 homozygotes: A Bayesian reanalysis of lecanemab and donanemab phase 3 results. Alzheimer's & Dementia: Translational Research & Clinical Interventions. 2025;11:e70083. DOI: 10.1002/trc2.70083. PMID: 40225236. PMCID: PMC11982174.

5. Shang J, Zhong S, Shang L, et al. Real-world application of lecanemab in early-stage Alzheimer's disease: a single-center prospective cohort analysis. Alzheimer's Research & Therapy. 2025. DOI: 10.1186/s13195-025-01886-5. PMID: 41250246. PMCID: PMC12625011.

Disclaimer: The image accompanying this article is for illustrative purposes only and does not depict actual experimental results, data, or biological mechanisms.