Neuroimaging Markers Associated With Recurrent Stroke in Intracerebral Hemorrhage and Atrial Fibrillation: Secondary Analysis of PRESTIGE-AF.

Abstract

Anticoagulation in patients with intracerebral hemorrhage (ICH) and atrial fibrillation reduces the risk of ischemic stroke (IS), but also potentially causes an excess risk of recurrent ICH. In this context, we assessed the role of neuroimaging in identifying patients with particular susceptibility to IS or ICH, potentially allowing for individualized risk stratification. In PRESTIGE-AF, a prospective, randomized clinical trial conducted across 75 hospitals in 6 European countries, participants with spontaneous ICH and atrial fibrillation were randomized to treatment with direct oral anticoagulants or no anticoagulation. Within a prespecified imaging subanalysis, we centrally assessed brain CT and MRI scans at baseline using established scales and definitions. We performed Cox regression analyses to investigate associations of neuroimaging findings with recurrent ICH and IS. PRESTIGE-AF included 319 patients, and 313 had neuroimaging of sufficient quality (median age 79 years, 35.5% female). MRI was available in 170 patients (54.3%). During a median follow-up of 1.4 years, 13 patients had recurrent ICH and 22 patients had IS. ICH recurrence was not associated with lobar vs nonlobar hematoma location or the overall category of probable cerebral amyloid angiopathy (<i>p</i> > 0.2). However, there was an increased risk of recurrent ICH in patients with cortical superficial siderosis (hazard ratio [HR] 7.7, 95% CI 1.4-42.2) and chronic intracerebral macrohemorrhages on MRI (HR 9.1, 95% CI 1.8-46.8). Patients with nonlobar ICH were at increased risk of IS (HR 9.1, 95% CI 1.2-67.7). Neuroimaging contributes to the identification of patients with ICH and atrial fibrillation at particularly high risk of recurrent ICH or IS, emphasizing its role in clinical decision making. MRI allows for the assessment of relevant neuroimaging markers that can aid the risk assessment in affected patients. Major study limitations include the modest number of outcome events, with confirmation of our findings needed in larger collaborations. ClinicalTrials.gov NCT03996772; first submitted June 21, 2019; first patient enrolled: May 31, 2019; available at clinicaltrials.gov/study/NCT03996772.