Imaging for anti-amyloid therapy eligibility: Comparing GRE and SWI in detecting microhemorrhages and superficial siderosis.

INTRODUCTION: Anti-amyloid therapies such as lecanemab require magnetic resonance imaging (MRI) screening to exclude patients with > 4 gradient-recalled echo (GRE)-detected microhemorrhages or any superficial siderosis (SS). Although GRE and susceptibility-weighted imaging (SWI) are both accepted, they differ in detection performance. METHODS: We retrospectively analyzed 224 eligibility MRIs with paired GRE and SWI from a single academic health system. Six neuroradiologists assessed a 20-scan reliability subset; the remaining 204 scans were assigned to one rater for hemorrhagic lesion detection and MRI-based eligibility classification. Interobserver reliability was calculated using intraclass correlation coefficients (ICCs), and sequence differences were evaluated using non-parametric tests. RESULTS: SWI detected more microhemorrhages than GRE (mean 1.24 vs. 0.67, p < 0.0001), identified eight additional SS cases, and showed higher interobserver agreement (ICC 0.951 vs. 0.764, p < 0.01). SWI classified more patients as MRI ineligible than GRE (26/204 [12.7%] vs. 11/204 [5.4%]). DISCUSSION: SWI identifies more exclusionary hemorrhagic findings and demonstrates greater reproducibility than GRE.