In a multi-year study of 81 adults, conducted by the Kentucky Neuroscience Institute, the CANS-MCI has been shown to be sensitive to heightened CSF levels of Aβ and tau levels.
The study examined cerebrospinal fluid (CSF) amyloid β 1-42 (Aβ) and tau levels and performance on a computerized self-administered test battery, the Computer-Administered Neuropsychological Screen for MCI (CANS-MCI).
Background: There is evidence of slight impairments on memory testing in the earliest, preclinical stages of the Alzheimer’s disease (AD), these individuals still perform well within healthy ranges. Computerized assessments can capture item-level response times. The present study explored within-session learning in preclinical AD by operationalizing learning as item response time on a self-administered, computerized object identification test. That is, participants should respond faster within the session as they ‘learn’ the nature of the simple task.
Methods: CSF was collected from participants near to the time of completing the baseline Computer- Administered Neuropsychological Screen for MCI (CANS). CSF levels of Aβ (threshold = 250pg/ml) were used to group participants as elevated (Aβ+) and not elevated (Aβ-) brain amyloid. A linear mixed model (LMM) with object identification response time at each (correct) trial as the dependent variable and age, sex, education, depression, and trial as independent variables, a random slope of trial for participants. Slope for each participant was extracted and used to predict brain amyloid status and mean response time on the same task 6 months later (logistic regressions).
Results: Of 96 participants, 57 were Aβ- and 39 were Aβ+. The groups did not differ statistically in sex, education, depression, or prior CANS administrations, ps > 0.33. The Aβ+ group (76.6y) was slightly older than Aβ- (73.9y), p =0.05. The LMM indicated age (β = 0.02, p < 0.001) and depression (β = 0.03, p = 0.04) significantly slowed response time, and response time was significantly faster with successive trials (β = -0.04, p < 0.001). Age (β = -0.06, p = 0.06) and participant slopes for trial (β = -75.16, p = 0.01) correctly classified 36 Aβ- (63.2%) and 28 Aβ+ (71.8%) participants. For the Aβ+ group, follow-up response times varied little by slope, whereas Aβ- participants with steeper slopes had faster times at follow-up.
Conclusion: Aβ+participants exhibited significantly steeper learning slopes than Aβ-. This may reflect slight retrieval delays early in the task and subsequent hyperactivation. Learning slopes at baseline also predicted improved performance at follow-up, in line with existing research on within-session practice effects. Implications and future directions discussed.