Amyloid and tau may moderate practice effects in semantic and episodic memory in a cognitively unimpaired at-risk sample

Previous studies in pre-clinical Alzheimer's disease have found associations between amyloid and practice effects. Additionally, studies within our research group have found associations between practice effects and cognitive tests, including language-based measures, as well as associations between language-based measures and Alzheimer's disease biomarkers. The purpose of this study is to bridge the gaps between these areas to further understand how practice effects may or may not explain additional variance between biomarker groups and to expand on current literature by incorporating tau status into these models. Our study had three main aims: (1) determine which of our proposed operationalizations of practice effects performed best, (2) explore the impact of amyloid on practice effects and (3) explore the impact of combined amyloid and tau status on practice effects for language-based measures: the proper names composite from the Logical Memory story recall task, as well as the total score, animal fluency, and letter fluency tasks. Participants from the Wisconsin Registry for Alzheimer's Prevention study with amyloid positron emission tomography (PET) scans and item-level Logical Memory data were included in these analyses (<i>n</i> = 442); the Aim 3 subset, requiring both amyloid and tau PET scans, included <i>n</i> = 397. Linear mixed effects models were used to explore our aims; for Aim 1 we utilized Akaike information criteria (AIC) to determine which operationalization performed best and for Aims 2 and 3 we used an interaction of biomarker * practice to determine if biomarker status(es) moderated the impact of practice on our measures of interest. Comparing the base model AIC to the models including practice showed that inclusion improved model fit for all outcomes Proper names from Logical Memory and the total score of Logical Memory showed a moderating effect of amyloid status on practice effects. Sensitivity analyses indicated that it may be age that is driving the association, however. A similar pattern was seen upon testing amyloid/tau (A/T) status and practice effect moderation on our outcomes, such that participants who were A+T+ did not appear to benefit from practice as much as the A-T- participants did. Future studies should seek to tease apart the intertwined impacts of age, practice and cognitive decline.