Critical ReviewMedicine & Health

Alzheimer's Drugs Finally Work — But the Side Effects Define the Debate

Lecanemab's Phase 3 trial demonstrated a 1.75-point reduction in cognitive decline over 4 years—but with a 4.35x increase in ARIA risk compared to placebo. As donanemab gains FDA approval, the benefit-to-risk calculus is now a clinical reality that divides neurologists, patients, and payers.

By Sean K.S. Shin
This blog summarizes research trends based on published paper abstracts. Specific numbers or findings may contain inaccuracies. For scholarly rigor, always consult the original papers cited in each post.

For decades, Alzheimer's disease research was defined by failure. Drug after drug targeting amyloid plaques failed in clinical trials, leading many researchers to question the amyloid hypothesis itself. That era is over—not because the debate was resolved elegantly, but because clinical trial data forced a reckoning. Lecanemab's Phase 3 trial (Clarity AD), published in the New England Journal of Medicine by van Dyck et al., showed statistically significant slowing of cognitive decline. Donanemab received FDA approval in July 2025. Anti-amyloid antibodies work, at least by the standard of clinical endpoints.

But "works" is doing heavy lifting in that sentence. The central question has shifted from "can we slow Alzheimer's?" to "is the benefit worth the risk?"—and the answer depends on who is doing the calculation.

The Research Landscape

What the Trial Showed

The Clarity AD Phase 3 trial enrolled patients with early Alzheimer's disease—mild cognitive impairment or mild dementia with confirmed amyloid pathology. Participants received lecanemab (10 mg/kg biweekly intravenous infusion) or placebo over 18 months, with follow-up data now extending to 4 years.

The primary endpoint was change from baseline on the Clinical Dementia Rating–Sum of Boxes (CDR-SB), a composite measure of cognitive and functional decline. Lecanemab reduced decline by 1.75 points over 4 years compared to placebo. The drug also significantly reduced amyloid plaque burden, as measured by PET imaging—confirming that the mechanism of action (amyloid clearance) was engaged.

The Side Effect That Defines Everything

Amyloid-related imaging abnormalities (ARIA) emerged as the defining safety concern. ARIA encompasses two subtypes: ARIA-E (edema/effusion—fluid accumulation in the brain) and ARIA-H (hemorrhage—microbleeds or superficial siderosis). Most ARIA cases are radiologically detected and clinically asymptomatic, but a subset produces symptoms including headache, confusion, dizziness, and visual disturbance. Severe cases can be life-threatening.

The trial reported that ARIA risk was 4.35 times higher in the lecanemab group compared to placebo. APOE4 homozygotes—who carry two copies of the gene variant most strongly associated with Alzheimer's risk—face substantially higher ARIA rates. This creates a clinical irony: the patients at highest genetic risk for Alzheimer's are also at highest risk for the treatment's most dangerous side effect.

Donanemab: The Second Entry

Donanemab, developed by Eli Lilly, received FDA approval in July 2025 based on its own Phase 3 data (TRAILBLAZER-ALZ 2) showing similar cognitive benefits and similar ARIA risk profiles. The approval of two anti-amyloid antibodies within two years confirms that this drug class represents a genuine therapeutic advance—while also doubling the urgency of the benefit-risk question.

Critical Analysis: Claims and Evidence

<
ClaimSourceVerdict
Lecanemab reduced cognitive decline over 4 years (1.75-point CDR-SB difference)Van Dyck et al., Phase 3 trial✅ Supported — statistically significant primary endpoint
ARIA risk was 4.35x higher than placeboVan Dyck et al., safety data✅ Supported — consistent across trial populations
Donanemab received FDA approval (July 2025)FDA regulatory record✅ Confirmed
The clinical benefit is meaningful to patientsInterpretation of CDR-SB change⚠️ Debated — 1.75 points over 4 years may or may not be perceptible to patients and caregivers
APOE4 carriers face disproportionate ARIA riskSubgroup analyses✅ Supported — consistent finding across both drugs

The Debate: Who Decides What "Worth It" Means?

The clinical controversy is not about whether the drugs work in a statistical sense—they do. It is about whether the magnitude of benefit justifies the magnitude of risk, cost, and treatment burden.

The case for treatment: Alzheimer's is a progressive, fatal disease with no prior disease-modifying therapy. Even modest slowing of decline extends the period of independence. For patients and families, months of preserved function may be profoundly meaningful in ways that CDR-SB point differences fail to capture.

The case for caution: A 1.75-point difference on an 18-point scale over 4 years translates to roughly a 5-month delay in disease progression by some estimates. The treatment requires biweekly infusions, regular MRI monitoring for ARIA, and costs approximately $26,500 per year (lecanemab). ARIA monitoring alone adds significant healthcare system burden.

The equity dimension: These are infusion therapies requiring specialized centers, PET scans for amyloid confirmation, and serial MRI monitoring. Access is structurally limited to patients near major medical centers with insurance coverage. The Alzheimer's patients who stand to benefit most—early-stage, confirmed amyloid-positive—are also the hardest to identify in community practice.

Open Questions

  • Clinical meaningfulness: How should clinical meaningfulness be defined for Alzheimer's therapies? Statistical significance on a rating scale and meaningful impact on daily life are not the same thing. Patient-centered outcomes research is urgently needed.
  • Risk stratification: Can ARIA risk be predicted at the individual level beyond APOE4 status? Biomarkers that identify high-ARIA-risk patients before treatment would transform the benefit-risk calculation.
  • Combination therapy: Will anti-amyloid antibodies combine effectively with therapies targeting tau, neuroinflammation, or synaptic dysfunction? Amyloid clearance alone slows but does not stop progression.
  • Duration and discontinuation: What happens when treatment stops? Does amyloid reaccumulate? Is there an optimal treatment duration, or must patients continue indefinitely?
  • Health system readiness: The global Alzheimer's population exceeds 55 million. Even if only the early-stage, biomarker-confirmed subset is eligible, the infusion capacity, MRI monitoring infrastructure, and cost burden are staggering. Is any health system prepared?
  • What This Means

    The Alzheimer's drug debate is no longer about whether disease-modifying therapy is possible. It is possible. The debate is now about clinical judgment under uncertainty: how to weigh modest benefit against real risk for individual patients with a fatal disease and no alternatives. This is not a failure of science—it is the ordinary, difficult work of translating trial data into clinical practice.

    Neurologists, patients, families, and payers will reach different answers. That disagreement is appropriate given the evidence. What would be inappropriate is pretending the answer is simple.

    Explore related work through ORAA ResearchBrain.

    References (2)

    [1] van Dyck, C.H., et al. Lecanemab in Early Alzheimer's Disease. New England Journal of Medicine.
    van Dyck, C. H., Swanson, C. J., Aisen, P., Bateman, R. J., Chen, C., Gee, M., et al. (2023). Lecanemab in Early Alzheimer’s Disease. New England Journal of Medicine, 388(1), 9-21.

    Explore this topic deeper

    Search 290M+ papers, detect research gaps, and find what hasn't been studied yet.

    Click to remove unwanted keywords

    Search 7 keywords →