Risk of amyloid-related imaging abnormalities associated with anticoagulant therapy in patients with Alzheimer’s disease treated with anti-amyloid monoclonal antibodies: a systematic review and meta-analysis

Introduction

Anti-amyloid monoclonal antibodies have emerged as a disease-modifying treatment option in early Alzheimer’s disease (AD). In patients receiving anti-amyloid immunotherapy, Aβ removal-associated blood-brain barrier damage and perivascular inflammation may present as amyloid-related imaging abnormalities (ARIA) or intracerebral haemorrhage (ICH). Based on theoretical and empirical concerns that ARIA might be more common and more severe in patients receiving oral anticoagulants, current recommendations caution against the combination of oral anticoagulants and anti-amyloid antibodies.^{1 2} In the European Union (EU) and UK, anti-amyloid antibody therapies are not authorised for patients receiving anticoagulants.

Considering the high prevalence of cardiovascular comorbidities in patients with AD and the crucial role of anticoagulation in the prevention of ischaemic stroke in patients with atrial fibrillation, we conducted a systematic review and meta-analysis of the risk of ARIA associated with anticoagulant use in patients with AD receiving anti-amyloid therapies.

Methods

PubMed and Web of Science were last searched on 19 July 2025, for studies of anti-amyloid treatment for early AD. We included randomised and non-randomised studies reporting rates of ARIA or ICH stratified by anticoagulant use and excluded non-human data and single-case reports. We used stratified random-effects meta-analysis to estimate the risk of ARIA associated with anticoagulant use and to evaluate differences between anti-amyloid therapies.

This study was registered with the International Prospective Register of Systematic Reviews (PROSPERO, CRD420251088254) and follows the Cochrane Collaboration and Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Data that support the findings of this study are available from the corresponding authors.

Results

ARIA rates stratified by anticoagulant use were analysed from randomised trials of donanemab (TRAILBLAZER-ALZ and ALZ-2) and lecanemab (Clarity-AD), their open-label addendum/extension and a real-world memory-clinic study (figure 1A).³,⁵

Figure 1. (A) PRISMA flow diagram. The number of identified, screened, assessed and included reports, together with the number of reports excluded at each stage, is indicated. (B–C) Forest plots illustrating stratified meta-analyses of the association between oral anticoagulant use (OAC) and amyloid-related imaging abnormalities (ARIA) in patients with Alzheimer’s disease receiving anti-amyloid immunotherapy. Absolute numbers of patients with and without OAC and with (+) and without (−) ARIA-E (B)) and ARIA-H (C)) are indicated for each of the included studies. Subtotals for lecanemab and donanemab are indicated in bold face. Trial-specific risk ratios and 95% confidence intervals (CI) are illustrated by black squares and line segments, respectively. Drug-specific and overall risk ratios were estimated using random effect (RE) models and are illustrated by dark blue and light blue diamonds, respectively. Within-group heterogeneity is quantified by estimates of the variance of the effect sizes across studies (τ²) and the proportion of variation in observed effects not due to sampling error (I²). Different definitions of ARIA-H are labelled (a: any ARIA-H; b: ARIA-H without ARIA-E). AD, Alzheimer’s disease; ARIA-E, ARIA with oedema/effusion; ARIA-H, ARIA with microhaemorrhage/superficial siderosis; ICH, intracerebral haemorrhage; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

In total, 286 of 3837 participants received oral anticoagulants while undergoing anti-amyloid immunotherapy (lecanemab, 156/1806; donanemab, 130/2031). ARIA with oedema/effusion (ARIA-E) or ARIA with microhaemorrhage/superficial siderosis (ARIA-H) was reported in 14% and 20% of patients, respectively (lecanemab, 9% and 13%; donanemab, 21% and 28%). Meta-analysis showed similar rates of ARIA-E (14.3% vs 18.1%, pooled risk ratio (RR) 0.83; 95% CI 0.63 to 1.11) and ARIA-H (20.0% vs 23.6%, RR 0.86; 95% CI 0.68 to 1.09) in patients with and without anticoagulant use. There were no significant subgroup differences, with rates of ARIA-E and ARIA-H favouring anticoagulation over no-anticoagulation in patients treated with either lecanemab or donanemab (figure 1B,C).

Macrohaemorrhages occurred in four patients (all treated with lecanemab) receiving anticoagulants (1.4%) and eight patients (five lecanemab, three donanemab) not receiving anticoagulants (0.23%).

Discussion

The meta-analysis outcomes of four clinical trials and a real-world study did not support the claim that oral anticoagulant use is associated with an increased risk of ARIA-E or ARIA-H in patients receiving anti-amyloid monoclonal antibody therapy. Due to low numbers of ICHs, especially with donanemab, any excess ICH risk associated with anticoagulant use could not be estimated precisely. Our meta-analysis did not support the claim that anti-amyloid antibody therapies should be avoided in patients treated with oral anticoagulants.

Our study has limitations. Classes of anticoagulants and the use of other concomitant antithrombotic therapies were seldom reported. ARIA was analysed as reported and not centrally adjudicated, which could affect between-trial comparability. Since anticoagulation was an exclusion criterion or not reported in clinical trials of other anti-amyloid immunotherapies (aducanumab, bapineuzumab, crenezumab, gantenerumab, solanezumab), no stratified safety data were available for these agents, which limits generalisability. Similarly, stratification by other risk factors of ARIA—like APO-ε4 carriership, baseline number of microhaemorrhages and white matter abnormalities^{4 5}—was not reported, preventing the analysis of potential interactions between baseline cardiovascular risk factors and anticoagulant use. Finally, non-randomised exposure to anticoagulants increases the risk of selection bias and unmeasured confounding.

Despite providing reassuring safety signals, our findings should be interpreted cautiously given the small number of patients treated with both anti-amyloid antibodies and anticoagulants. To more precisely define risks and benefits of anti-amyloid therapies in this subpopulation, our data support the inclusion of patients with AD receiving anticoagulants in carefully designed clinical trials, possibly exploring lower antibody doses, modified titration regimes, extended treatment intervals and alternative strategies to remove amyloid.

This article does not represent the official viewpoints of the University Medical Center Hamburg-Eppendorf, McGill University or Peter O'Donnell Jr. Brain Institute.