TY - JOUR T1 - Multiple memory tests in midlife improve prediction of mild cognitive impairment JF - bioRxiv DO - 10.1101/584193 SP - 584193 AU - Daniel E. Gustavson AU - Mark Sanderson-Cimino AU - Jeremy A. Elman AU - Carol E. Franz AU - Matthew S. Panizzon AU - Amy J. Jak AU - Chandra A. Reynolds AU - Michael J. Lyons AU - William S. Kremen Y1 - 2019/01/01 UR - http://biorxiv.org/content/early/2019/03/21/584193.abstract N2 - Objective Biomarkers predict risk for Alzheimer’s disease, but the ability to predict when most people are still biomarker negative would foster earlier identification of risk. We hypothesized that memory testing—particularly combining multiple memory measures—in cognitively normal middle-aged adults would provide good prediction of 6-year progression to mild cognitive impairment (MCI). Our use of multiple measures contrasts with frequent efforts to identify the single best measure for prediction, an underlying assumption of which is that adding a suboptimal measure will worsen prediction.Method We examined how Logical Memory, Visual Reproductions, and California Verbal Learning Test (CVLT) predict 6-year progression to MCI in 849 men who were cognitively normal at baseline (mean age=55.69±2.45).Results CVLT learning trials was the best individual predictor of amnestic MCI (OR=4.75). However, a latent factor score incorporating 7 measures across the 3 memory tests provided the strongest prediction (OR=9.88). This OR compared favorably to biomarker prediction in a study of substantially older adults.Interpretation Use of a single best measure may appear to be time- and cost-effective, but 30 additional minutes of memory testing—even when it includes individually less predictive measures—provides a very worthwhile tradeoff for substantially improved prediction of progression to amnestic MCI. Early identification is important for potentially slowing Alzheimer’s disease progression, and these results indicate that neuropsychological tests can be very early sensitive indicators of risk at an age when few individuals will have yet become biomarker positive. ER -