Identifying factors affecting resilience in atrophy-based subtypes of vascular cognitive impairment

Abstract

Introduction: Vascular cognitive impairment (VCI) is heterogeneous in brain atrophy patterns, clinical symptoms, and possibly in the factors affecting resilience to symptoms. The objective of this study was to investigate the heterogeneity of VCI by estimating different atrophy-driven subtypes and use those for identifying resilience factors in each subtype. Methods: We used cross-sectional data from the Trace-VCI cohort comprising of memory-clinic patients with vascular brain injury on MRI (n=361 all-cause dementia, n=190 MCI, and n=188 SCD). White matter hyper-intensities (WMH) were segmented, and SLF toolbox was used to refill them on the MRIs for accurate parcellation. Freesurfer volumes were used for identifying subtypes with non-negative matrix factorization. Subtype-specific pseudo-timelines of progression were estimated using a previously validated discriminative event-based model. Severity of atrophy (SA) in patients was estimated using cross-validation based on their position on the pseudo-timeline. Using linear-regression, cognition and disability (MMSE, Global deterioration scale, disability assessment for dementia) were modelled to be dependent on SA and its interactions with genetic factors, vascular markers and risk-factors, and co-pathology independently (variables in Figure-3). Resilience factors were identified by testing if the model with the interaction explains the symptoms significantly more than the one without. Results: The algorithm identified three atrophy-based VCI subtypes: Frontal, Subcortical/Temporal, and Parietal subtype. Their prevalence, vascular and clinical presentations are summarized in Figure-1. The pseudo-timelines of atrophy are shown in Figure-2. SA's interaction with education positively influenced cognition in all subtypes, but negatively influenced disability in subcortical/temporal subtype. In frontal and parietal subtypes, APOE ε4, AD co-pathology resulted in more SA without worsening symptoms. SA's interaction with smoking and microbleeds negatively influenced disability. In the subcortical/temporal subtype, SA's interaction with WMH, lacunes, infarcts negatively impacted disability. In parietal subtype, men have more cognitive resilience than women. SA's interaction with hypercholesterolemia, and smoking were significantly negative for cognition. Lacunes were associated with more SA without affecting cognition. Figure-3 summarizes the interactions for all subtypes. Discussion: We identified three atrophy-based VCI subtypes where the risk-factors have different influence on atrophy and symptoms highlighting differences in resilience. These results could aid in prognosis and in personalizing patients’ intervention strategy.