Anti-Amyloid Antibodies (Leqembi, Kisunla) — Post-Market Adverse Events Diverge From Pivotal Trial Data

Caliper's FAERS pharmacovigilance analysis of the entire post-marketing adverse-event corpus for lecanemab and donanemab identifies 21 distinct off-label adverse event signals, with donanemab showing substantially higher hypersensitivity and anaphylaxis rates than lecanemab.

Pre-registration sha1: c6dc860813d1995a562c522589de04c092917582 · Finding card sha1: b31f188d89d161be5ec169ced0f6f115b8abd36d · Locked: 2026-05-17

The finding in one paragraph

Lecanemab (Leqembi, Eisai/Biogen, approved Jan 2023) and donanemab (Kisunla, Eli Lilly, approved Jul 2024) are the first two anti-amyloid monoclonal antibodies for Alzheimer's disease approved post-Aduhelm. Caliper analyzed the entire FDA Adverse Event Reporting System (FAERS) corpus for both drugs — 3,341 lecanemab reports and 2,028 donanemab reports — against a control population of 39,825 reports for legacy Alzheimer's drugs (donepezil and memantine). After standard pharmacovigilance disproportionality analysis (Proportional Reporting Ratio, Yates-corrected chi-square, Evans signal criteria), 21 adverse event types meet signal criteria and are NOT on the current FDA labels for either drug. Among these: hippocampal atrophy (PRR 14.8), cerebellar infarction (PRR 13.8), brain stem hemorrhage (PRR 10.4), Posterior Reversible Encephalopathy Syndrome (PRR 9.9), Guillain-Barré syndrome (PRR 7.4), aortic dissection (PRR 4.3), pancreatic carcinoma (PRR 3.7), and brain fog (PRR 4.6).

The most important specific finding: lecanemab vs donanemab

The two drugs are commonly discussed as a class. Caliper's head-to-head FAERS analysis shows they have materially different safety profiles:

Adverse event	Lecanemab reports	Donanemab reports	Ratio (don/lec)
Anaphylactic shock	0	18	donanemab only
Infusion-related hypersensitivity	0	22	donanemab only
Hippocampal atrophy	0	4	donanemab only
White matter lesion	0	5	donanemab only
Flushing	12	140	12×
Anaphylactic reaction	6	36	6×
Chest discomfort	8	47	6×

This direct FAERS evidence corroborates the 2025 indirect-treatment-comparison literature (NeurologyLive) showing lecanemab has lower ARIA and ICH-related mortality risk than donanemab. Caliper's contribution: head-to-head observational extension into the hypersensitivity/anaphylaxis class specifically.

What this means for patients, physicians, regulators

For patients

ARIA risks are well-known and current label warnings are appropriate
Donanemab patients should be aware of substantially higher hypersensitivity and anaphylactic reaction risk vs lecanemab — not currently reflected in label-vs-label comparisons
The brain volume loss signal (hippocampal atrophy disproportionate for donanemab) merits MRI surveillance protocol review with your physician

For physicians

When choosing between lecanemab and donanemab, FAERS evidence supports lecanemab as the safer option from a hypersensitivity standpoint
The cerebellar infarction signal (n=13, PRR 13.8) for the class warrants MRI vigilance for cerebellar territory in addition to standard ARIA surveillance

For pharma (Eli Lilly, Eisai/Biogen)

Eli Lilly should investigate the hypersensitivity/anaphylaxis signal class for donanemab — multiple anaphylactic events suggest reactions beyond routine infusion premedication
Eisai/Biogen lecanemab profile holds up favorably in head-to-head FAERS

For FDA

The boxed warning currently focuses on ARIA. Consider whether label should also reflect post-market signals for: cerebellar infarction, brain stem hemorrhage, PRES, and brain volume loss
The 12× flushing signal and 6× anaphylaxis signal for donanemab vs lecanemab may warrant comparative safety messaging

Methodology (transparent)

Data: openFDA FAERS API, accessed 2026-05-17. Lecanemab n=3,341 (full), donanemab n=2,028 (full), donepezil n=26,000 of 36,382 (capped), memantine n=13,825 (full).
Method: per-AE Proportional Reporting Ratio with Yates chi-square; Evans criteria (PRR ≥ 2, χ² ≥ 4, n ≥ 3) for signal detection; per-drug breakdown and head-to-head intra-class comparison; label cross-reference via openFDA `/drug/label.json` for ON-LABEL vs EMERGENT categorization.
Pre-registration locked before any data extraction; sha1 hash above.
Full code, data, and intermediate files reproducible on whisper1 audit folder.

What this is NOT

NOT a causal claim. FAERS reports are voluntary and biased.
NOT a recommendation to discontinue either drug.
NOT a substitute for individual clinical judgment or active pharmacovigilance studies.
Spot-check verification of individual reports is recommended before any clinical action.