Investigating the temporal pattern of neuroimaging-based brain age estimation as a biomarker for Alzheimer's disease related neurodegeneration

The brain age gap (BAG) has been proposed as an Alzheimer's Disease (AD) biomarker and represents the difference between one's estimated brain age and chronological age. As AD-related neurodegeneration precedes clinical diagnosis, a longitudinal understanding of the BAG's temporal pattern is required. This study investigated how the BAG's trajectory varies by subject characteristics using data from the Alzheimer's Disease Neuroimaging Initiative. Brain age was estimated using a support vector regression model trained with multimodal features of cortical thickness via magnetic resonance imaging, and standardized uptake value ratios via fluorodeoxyglucose positron emission tomography data. Using multilevel modeling, the temporal pattern of the BAG was examined between participants given characteristics including clinical outcome, gender, and APOEɛ4 carriership. Performance of the brain age prediction model was comparable to past studies (mean absolute error = 3.71 years) and was observed to have a linear increasing trajectory across an 8-year period. Additionally, the BAG of individuals with progressive mild cognitive impairment increased at a significantly faster rate across time compared to stable groups, where this effect was stronger in females. This accelerated BAG trajectory was not moderated by AD diagnosis time. Our observations strengthen support for the BAG as a potential biomarker of AD by prioritizing a longitudinal design and interpretation, as opposed to previously reported cross-sectional evidence.