Errors and repetitions on tests of language, verbal learning, and memory in cognitively unimpaired adults at increased risk for Alzheimer’s disease

Abstract

AbstractBackgroundErrors and repetitions on verbal fluency and list learning tasks have been found to predict progression to MCI and differentiate between healthy aging, MCI, and dementia. This study examined whether errors and/or repetitions on verbal measures were cross‐sectionally associated with amyloid pathology in baseline cognitively unimpaired (CU) adults, and provided additional information beyond traditional total score measures.MethodWe included 437 participants from the Wisconsin Registry for Alzheimer’s Prevention who completed comprehensive neuropsychological testing and amyloid (Ab) positron emission tomography (PET) scans using Pittsburgh Compound‐B (PiB). Ab(+/‐) status was determined using an established threshold average distribution volume ratio(DVR) >1.19 across several brain regions. From the most recent testing visit, we extracted the total number of errors and repetitions from the Rey Auditory Verbal‐Learning Test (AVLT), animal fluency, and letter fluency; these were summed to form error and repetitions composites. We ran ANCOVAs with Ab(+/‐) as predictor; age, sex, and WRAT‐3 reading were covariates. Cohen’s D effect sizes characterized strength of relationships, where higher errors = negative effect sizes, and higher total scores = positive effect sizes. Sensitivity analyses examined errors and total scores from individual tests, and removal of participants who progressed to MCI.ResultSample characteristics are presented in Table1. Relationships between error, repetition and total score variables are shown in Figure1. Ab(+/‐) was a significant predictor of the error composite (p = .001) but not the repetition composite. Ab(+/‐) was significantly associated with AVLT learning total (p = .01), but not with animal or letter fluency totals. Secondary analyses revealed that AVLT errors alone were associated with Ab(+/‐) (p<.0001); effect sizes were greatest for AVLT errors (d = ‐.46) than for the error composite (d = ‐.39) or for AVLT total score (d = .39) (Figure 2). Removal of impaired participants yielded similar effect sizes.ConclusionFindings from this CU sample suggest that AVLT errors may be more sensitive to early AD pathology than traditional total score measures. Errors on list‐learning may represent other cognitive and/or sensory processes (e.g., auditory processing, hearing, attention, inhibition) that are sensitive to AD but are not represented in total scores. Future research includes examining longitudinal and additional error composites’ relationships with AD pathology to develop more sensitive cognitive measures.