A - 18 Predicting Incident Amnestic Mild Cognitive Impairment and Alzheimer's Disease with a Computerized Neuropsychological Assessment Device: Comparative Clinical Utility.

Abstract

Computerized neuropsychological assessment devices (CNADs) offer improved accessibility to screen for neurodegeneration, although their clinical utility is yet to be established. We evaluated the efficacy of a CNAD to predict incident amnestic mild cognitive impairment and Alzheimer's disease (aMCI/ad) compared to a conventional paper-and-pencil screening tool and PET amyloid beta (AB). Data were collected from the longitudinal observational Alzheimer's Disease Neuroimaging Initiative 3. Participants were cognitively normal at baseline (N = 315, mean age = 72.9 +/- 7.1, 59.4% female, 16.8 +/- 2.3years of education, 91.4% White, 33.7% APOE4+). Over four years, 26 (8.3%) participants converted to aMCI and 3 (1.0%) individuals developed ad. Prognostic validity was compared between three measures assessed at baseline. Computerized visual episodic memory scores were evaluated using the One Card Learning (OCL) test. Conventional screening was conducted using the Montreal Cognitive Assessment (MoCA). PET AB+ was quantified as ≥2 SD whole cerebellum referenced region standardized uptake value ratios. Area under the curve (AUC) analyses using logistic regression were adjusted for age, sex, education, race, and APOE4+ status. OCL accuracy yielded AUC = 0.67, p = 0.003, 95% CI [0.58, 0.76]. Total MoCA score demonstrated AUC = 0.75, p < 0.001, 95% CI [0.67, 0.83]. PET AB+ produced AUC = 0.68, p = 0.001, 95% CI [0.58, 0.79]. When assessed at baseline, the MoCA provided the greatest clinical utility to predict incident aMCI/ad. Baseline OCL accuracy and PET AB+ were similarly effective for determining future cognitive decline. Further research is needed to examine the utility of CNADs before they are integrated into clinical practice.