Objective: To identify the most accurate cognitive measures in discriminating between individuals with presymptomatic AD and individuals who remained nondemented.
Methods: During a 10-year prospective community study, 120 nondemented subjects completed a battery of standard cognitive tests and clinically manifested AD 1.5 years later. Performance on each of 16 cognitive tests was compared between these 120 presymptomatic cases and 483 controls who remained nondemented over the 10-year follow-up period. The area under the receiver operating characteristic (AUC) curve for each test was used to measure its accuracy of discrimination between cases and controls.
Results: Among the 16 neuropsychological tests, Word List Delayed Recall discriminated best between cases and controls (AUC = 0.806), followed by the Word List 3rd Learning Trial (0.787), Word List 1st Learning Trial (0.774), and Trail-making Test B (0.773), compared to the Mini-Mental State Examination (MMSE) (0.726). Both Word List Delayed Recall and Word List 3rd Learning Trial were significantly more accurate than the MMSE. The combination of Word List Delayed Recall and Trail-making Test B comprised the optimal set of cognitive measures, with the highest AUC (0.852).
Conclusion: Measures of delayed recall and executive functions were the best discriminators between those who would manifest AD 1.5 years later and those who would remain nondemented. These findings are relevant for the early detection of AD and, therefore, for prevention and early intervention trials. Executive dysfunction may be a subtle manifestation of incipient AD, along with memory dysfunction.