Preliminary Validation of a Structural Magnetic Resonance Imaging Metric for Tracking Dementia-Related Neurodegeneration and Future Decline

Abstract

AbstractCurrent Alzheimer’s disease (AD) research has a major focus on validating and discovering noninvasive biomarkers that can detect AD, benchmark disease severity, and aid in testing the efficacy of interventions. Structural magnetic resonance imaging (sMRI) is a well-validated tool used in diagnosis and for monitoring disease progression in AD. Much of the sMRI literature centers around hippocampal and other medial temporal lobe structure atrophy, which are strongly associated with cognition and diagnosis. Because atrophy patterns are complex and vary by patient, researchers have made efforts to condense more brain information into validated metrics. Many of these methods use machine learning (ML), which can be difficult to interpret clinically, hampering clinical adoption. Here, we introduce a practical, clinically meaningful and interpretable index which we call an “AD-NeuroScore.” Our approach is automated and uses multiple regional brain volumes associated with cognitive decline. We used a modified Euclidean inspired distance function to calculate the differences between each participant and a cognitively normal (CN) older adult template, adjusting for intracranial volume, age, sex, and scanner model. Here we report validation results, including sensitivity to diagnosis (CN, mild cognitive impairment (MCI), and AD) and disease severity (Clinical Dementia Rating Scale Sum of Boxes (CDR-SB), Mini Mental State Exam (MMSE), and Alzheimer’s Disease Assessment Scale-Cognitive Subscale (ADAS-11) in 929 older adults (mean age=72.7 years, SD=6.3, Range=55.1-91.5, 50% Female) drawn from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) study. To determine if AD-NeuroScore might be predictive of disease progression, we assessed the relationship between the calculated AD-NeuroScore at baseline and change in both diagnosis and disease severity scores at 12, 24, 36, and 48-months. We performed additional validation in all analyses, benchmarking AD-NeuroScore against adjusted hippocampal volume (AHV). We found that AD-NeuroScore was significantly associated with diagnosis and all disease severity scores at baseline. Associations between AD-NeuroScore and disease severity (CDR-SB and ADAS-11) were significantly stronger than with AHV. Baseline AD-NeuroScore was also associated with change in diagnosis and changes in disease severity scores at all time points. Performance was equivalent, or in some cases superior, to AHV. These early validation results suggest that AD-NeuroScore has the potential to be a clinically meaningful biomarker for dementia.