Neuropsychological Decline Improves Prediction of Dementia Beyond Alzheimer’s Disease Biomarker and Mild Cognitive Impairment Diagnoses

BACKGROUND: A clinical diagnosis of cognitive impairment is traditionally based on a single cognitive exam, but serial cognitive testing can be sensitive to subtle cognitive changes in asymptomatic individuals and inform cognitive trajectory. OBJECTIVE: We evaluated the prognostic utility of identifying longitudinal neuropsychological decline along with single cognitive exam and Alzheimer’s disease (AD) cerebrospinal fluid (CSF) biomarkers in predicting dementia. We also examined brain volumetric differences based on decline trajectories. METHOD: Regression models quantified 12-month neuropsychological decline relative to normative expectations among non-demented older adults (N = 1,074). Progression to dementia over follow-up (18-120 months) was diagnosed using independent modes of assessment. RESULTS: In Cox regression models controlling for age, sex, education, apolipoprotein E4, and baseline cognitive diagnosis, neuropsychological decline predicted increased dementia risk, χ(2) = 69.861, p < 0.001, odds ratio = 2.841, even after correction for CSF biomarkers (amyloid-β, phosphorylated tau, total tau), χ(2) = 26.365, p < 0.001, odds ratio = 2.283. Voxel-based morphometry analysis indicated smaller hippocampal and medial temporal volume in participants with neuropsychological decline. CONCLUSIONS: Longitudinal diagnosis of neuropsychological decline improved prognostic accuracy beyond single cognitive exam diagnoses and AD CSF biomarkers, even in asymptomatic older adults. Older adults with a trajectory of neuropsychological decline exhibit smaller medial temporal and hippocampal brain volume. Longitudinal diagnostic approaches may benefit selection and randomization procedures for AD clinical trials in asymptomatic individuals.