Anticoagulation in Intracerebral Hemorrhage Survivors with Atrial Fibrillation and Imaging Features of Cerebral Amyloid Angiopathy and Small Vessel Disease

Yale University, New Haven CT

Investigators

Abstract

PROJECT SUMMARY Non-traumatic intracerebral hemorrhage (ICH) is a devasting type of stroke caused by cerebral small vessel disease (CSVD). One-fourth of ICH patients have atrial fibrillation (AF), an important risk factor for ischemic stroke. Although anticoagulation therapy alleviates the risk of ischemic stroke in patients with AF, it can increase the risk of ICH. Observational data including our own, indicate that among ICH survivors, CSVD features such as microbleeds and white matter hyperintensities independently increase the risk of ischemic stroke, recurrent ICH, and poor functional outcomes. Whether CSVD similarly influences the net clinical benefit of anticoagulation in ICH survivors with AF represents a major knowledge gap. Therefore, the objective of this proposal is to examine the heterogeneity of anticoagulation treatment effect by CSVD phenotype and develop risk stratification tools to help guide anticoagulation therapy in ICH survivors with AF. Our central hypothesis is that anticoagulation is associated neutral effect or harm in patients with cerebral amyloid angiopathy and with net clinical benefit in patients with deep perforator arteriopathy. We will leverage prospective clinical and neuroimaging data from 1900 patients enrolled in ASPIRE and ENRICH-AF, two clinical trials evaluating the safety and efficacy of anticoagulation in ICH survivors with AF. Using baseline magnetic resonance imaging scans, we will ascertain individual CSVD features and adjudicate CSVD phenotypes. Aim 1 will evaluate if the four validated CSVD phenotypes (cerebral amyloid angiopathy, deep perforator arteriopathy, mixed, and undetermined) modify and mediate the association between anticoagulation and the outcomes of recurrent ICH (1A), major hemorrhage (1B), net clinical benefit of any stroke (1C), and good functional outcome defined by a modified Rankin score of 0-2 (1D). Aim 2 will develop a risk-stratification score for recurrent ICH using CSVD, demographic and clinical data from ENRICH-AF, which will then be tested in ASPIRE as an effect modifier of the association of anticoagulation therapy with recurrent ICH (2A), major hemorrhage (2B), net clinical benefit (2C), and good functional outcome (2D). Aim 3 will leverage deep learning models to develop and validate automated tools for the ascertainment of CSVD phenotypes, and subsequently evaluate if these automatically determined CSVD phenotypes modify the association of anticoagulation with recurrent ICH (3A), major hemorrhage (3B), net benefit (3C), and functional outcome (3D). This automated image analysis pipeline will be trained in ENRICH- AF and externally validated in ASPIRE. Impact: Regardless of the direction of the parent trials, successful completion of these aims will help identify subpopulations of ICH patients with AF who may experience benefit or harm from anticoagulation, taking the field one step closer to personalized medicine. The proposed research will have important multiplicative effects by generating new research and clinical tools, including a novel risk- stratification score, automated CSVD quantification tools, and the largest harmonized dataset of clinical trials evaluating anticoagulation in ICH survivors with AF assembled to date.