What are the rates and clinical consequences of ARIA in real‑world use of lecanemab and donanemab?
Executive summary
Real‑world reports and pooled analyses show that amyloid‑related imaging abnormalities (ARIA) are substantially more common with anti‑amyloid monoclonal antibodies than with placebo, and that lecanemab tends to generate lower ARIA rates and fewer severe hemorrhagic events than donanemab in both indirect comparisons and emerging clinical‑use series [1] [2] [3]. ARIA most often appears in the first year, is associated with APOE‑ε4 dose, is usually manageable with monitoring and temporary treatment interruption, but in trial programs it has caused disabling hemorrhage and a small number of deaths—risks that inform current MRI surveillance and dosing changes [4] [5] [6] [7].
1. What the data show about how often ARIA occurs in practice
Real‑world surveillance and pooled safety reviews place ARIA incidence well above placebo background and broadly in line with trial experience: a pooled review reported overall ARIA‑E in 14% and ARIA‑H in 20% across lecanemab and donanemab cohorts, with drug‑specific real‑world rates cited as roughly 9% ARIA‑E/13% ARIA‑H for lecanemab and 21% ARIA‑E/28% ARIA‑H for donanemab in the datasets examined [8]. Industry‑led two‑year U.S. center series for lecanemab reported that discontinuations directly attributable to ARIA were uncommon (ARIA‑E in 1.1%, ARIA‑H in 1.1%, combined ARIA in 0.6% of patients), and that most treated patients continued on therapy (87.4% at the reporting timepoint) [9] [10]. Meta‑analytic work combining randomized and real‑world studies quantifies the class effect: anti‑Aβ antibodies increased ARIA risk roughly fourfold versus control, with an especially large relative increase for ARIA‑E [11] [1].
2. How lecanemab and donanemab compare on ARIA frequency
Multiple analyses converge on a consistent pattern: donanemab has produced higher ARIA rates in trials and observational series than lecanemab. Phase‑3 and program data cited ARIA‑E rates near 12–17% for lecanemab and substantially higher ARIA‑E/ARIA‑H proportions for donanemab (trial summaries report donanemab ARIA‑E ≈24% and ARIA‑H ≈31.4% in TRAILBLAZER‑ALZ2) [12] [6]. Indirect treatment comparisons and network meta‑analyses presented at recent conferences and in independent summaries found a statistically significant lower risk of any ARIA, ARIA‑E and ARIA‑H with lecanemab versus donanemab, and reported lower concurrent ARIA/ICH‑related mortality for lecanemab in those comparisons [2] [3] [13].
3. Clinical consequences and severity: beyond imaging to outcomes
Most ARIA events are radiographic and asymptomatic or transiently symptomatic, and are managed by pausing therapy and repeating MRI surveillance per practice guidance [7]. However, the class carries real clinical danger: trial programs documented intracerebral hemorrhage and stroke‑like presentations, and reporting to date attributes a small number of deaths to ARIA in the donanemab program (three deaths cited across donanemab trials) and serious hemorrhagic events within both drug programs [6]. Industry and academic reviewers stress that these are low‑frequency but high‑impact events that drive rigorous pre‑treatment MRI screening, APOE genotyping recommendations, and modified dosing/titration strategies to reduce risk [5] [7] [13].
4. Risk modifiers, monitoring and evolving safety responses
APOE‑ε4 genotype strongly modifies ARIA risk—evidence supports a gene‑dose effect for lecanemab in particular and underlies recommendations for genotyping and closer monitoring of homozygotes [5]. Concomitant anticoagulation, while a theoretical concern, has not shown a clear signal of raising ARIA rates in pooled analyses to date, though the low number of ICH events limits precision of that finding [8]. Regulators and companies have acted: donanemab’s label and titration recommendations were revised to slow dosing and reduce ARIA risk, and clinicians follow MRI surveillance schedules outlined in appropriate‑use documents for lecanemab [13] [7].
5. What remains uncertain and what the data limitations are
Real‑world evidence is growing but fragmented: many reports come from industry‑sponsored registries or single‑center series with limited follow‑up, and indirect comparisons rely on trial‑level anchors rather than head‑to‑head randomized data [9] [10] [3]. That means precise absolute risk estimates, long‑term outcomes after ARIA, and subgroup risks across diverse clinical populations remain incompletely characterized; authoritative large‑scale registry and independent post‑marketing surveillance will be needed to refine those numbers [4] [1].