Title: Associations Between Episodic Memory and Hippocampal Volume in Late Adulthood

Abstract: There are various ways of conceptualizing and assessing episodic memory, but different tasks designed to assess episodic memory are only moderately correlated with each other, suggesting that episodic memory might not be a unitary construct. Further, various episodic memory tasks exhibit disproportional task demands on the hippocampus and differentially reflect hippocampal volume degeneration – one of the strongest predictors of Alzheimer’s disease. Therefore, it is unclear if variation in performance on episodic memory measures is a meaningful indicator of risk for developing Alzheimer’s disease. This study established a structural equation model to examine the covariance structure and distinctiveness of tasks that have been traditionally used as measures of episodic memory and assessed whether these relate differently to hippocampal volume. We examined 648 older adults (M= 69.88) with 16.32 years of education. Episodic memory was assessed using Logical Memory, Paired Associates, HVLT-R, MoCA, Picture Sequence Memory Test, BVMT-R, and Cohen’s Relational Memory Test. Automated Segmentation of Hippocampal Subfields was used to segment the hippocampus. A confirmatory factor analysis was used with residual covariances included and loadings freely estimated. Hierarchical regression models were used to test the associations between the observed factors of episodic memory and hippocampal volume. A model with three first-order subfactors (verbal immediate recall, verbal delayed recall, and visuospatial) derived from a second-order general episodic memory domain factor had the best model fit. The verbal delayed recall subfactor explained the most variance in total and left hippocampal volume, whereas the verbal immediate recall subfactor explained the most variance in right hippocampal volume. In addition, the verbal delayed recall subfactor explained the most variance in CA1 and CA3 volume, the verbal immediate recall subfactor explained the most variance in entorhinal cortex volume, and the visuospatial subfactor explained the most variance in subiculum volume. However, the amount of variance in hippocampal volume explained was similar across all subfactors. No subfactors were significantly associated with CA2 or dentate gyrus volume. Our results suggest that traditional episodic memory tasks are measuring the same overarching construct, but different task conditions are tapping into different complex processes associated with episodic memory. Further, performance across the observed factors only accounted for a small portion of the variance in hippocampal volume, suggesting that hippocampal volume might not be a strong marker of episodic memory ability before clinically observable cognitive deficits are present. Lastly, findings from our study suggest that different hippocampal subfields are not uniformly involved in managing and supporting episodic memory, and they may be preferentially important for various processes.