Baseline brain volumes predict cognitive decline more robustly than atrophy rates: Evidence for brain reserve

The concept of brain reserve offers a framework for understanding the mechanisms of cognitive resilience in later life. This study investigated whether a static measure of brain structure, a proxy for accumulated reserve, or a dynamic measure of atrophy is a more powerful predictor of future cognitive changes. Using Alzheimer's disease (AD) as a model of accelerated cognitive aging, we investigated this question across the neurocognitive continuum. We analyzed longitudinal structural MRI and neuropsychological data from 75 Alzheimer's Disease Neuroimaging Initiative (ADNI) participants (Healthy, MCI, Dementia) using LASSO regularization and logistic regression to identify associations with cognitive change and model MCI conversion. Baseline brain volumes were robust predictors of cognitive decline, significantly outperforming short-term volumetric changes (atrophy rate). Specifically, larger lateral ventricles and smaller hippocampal and thalamic volumes, particularly in the anterior and medial thalamic nuclei, were associated with greater impairment in memory and executive function. Importantly, while APOE ϵ4 carrier status was high across the MCI group (∼56%), it did not differentiate converters from non-converters, indicating that genetic risk alone does not account for divergent clinical trajectories. Our model, based on these baseline volumes, predicted MCI conversion with good accuracy (AUC = .82). These findings highlight the importance of structural brain reserve, as a proxy for the resilience of integrated neural networks against neurodegeneration. Our results underscore the clinical utility of a single baseline MRI scan for identifying individuals at risk and advancing a network-based understanding of neurodegeneration.